Corporate Portfolio Management within Japanese Diversified Trading & Investment Companies
- What Role Does Real Estate Play? by
Takanori Ono
B.A., Economics, 2005
Kyoto University
Submitted to the Program in Real Estate Development in Conjunction with the Center for Real
Estate in Partial Fulfillment of the Requirements for the Degree of Master of Science in Real Estate
Development
at the
Massachusetts Institute of Technology
September, 2012
©2012 Takanori Ono
All rights reserved
The author hereby grants to MIT permission to reproduce and to distribute publicly paper and
electronic copies of this thesis document in whole or in part in any medium now known or hereafter
created.
Signature of Author
Center for Real Estate
July 30, 2012
Certified by
David Geltner
Professor of Real Estate Finance, Department of Urban
Studies and Planning
Thesis Supervisor
Accepted by
David Geltner
Chair, MSRED Committee, Interdepartmental
Degree Program in
Real Estate Development
2"
"
Corporate Portfolio Management within Japanese Diversified Trading & Investment Companies
- What Role Does Real Estate Play? by
Takanori Ono
Submitted to the Program in Real Estate Development in Conjunction with the Center for Real
Estate on July 30, 2012 in Partial Fulfillment of the Requirements for the Degree of Master of
Science in Real Estate Development
ABSTRACT
This paper discusses possible optimal corporate portfolio composition for Japanese trading and
investment firms from stakeholders’ (specifically shareholders and employees) value maximization
perspective. Based on the historical returns of diversified business units of 4 subject companies,
performances of individual business units and three portfolios (current, tangency, and “suboptimal”) are
analyzed and compared. The study suggests adjusting suboptimal portfolio composition based on each
business unit’s systematic risk and excess market return relative to its systematic risk and industry
average. A firm also needs consideration on how the composition adjustment would affect diversification
benefits the firm now enjoys and also on its overall management strategy.
Key words: corporate portfolio management, diversification, stakeholder theory, portfolio theory, CAPM,
Index model, accounting beta, Jensen’s Alpha, Treynor ratio, multi-factor model
Thesis Supervisor: David Geltner
Title: Professor of Real Estate Finance
3"
"
4"
"
ACKNOWLEDGEMENT
First and foremost, I’d like to thank Professor David Geltner, my advisor. He has always
inspired me intellectually, and I’ve learned a lot from him. Without his insights and constructive
advice, I couldn’t have completed this research.
I would also like to express my gratitude to faculty and administrative staff of MIT
Center for Real Estate for their continuous support and guidance. They have tremendously
leveraged my learning experience.
My special thanks are extended to my classmates. They have always inspired me cheered
me up. Especially, they have magically turned my potentially painful graduate school life into
meaningful and unforgettable one by their wit and personalities.
I also wish to acknowledge the support and assistance offered by Sumitomo Corporation,
my sponsor. I am particularly grateful for the assistance given by members of Strategic Real
Estate Business Department. I cannot thank them enough for giving me this opportunity.
Last but not least, I’d like to express my sincere appreciation to my parents, family and
friends. They have always believed in me and unconditionally supported me. Because I knew
they were always at my back, I could challenge myself to step out of my comfort zone.
5"
"
TABLE OF CONTENTS
1.
2.
3.
Introduction
1.1.
Research Motivation
1.2.
Approach and Structure
11
1.3.
Overview of Diversified Trading & Investment Company
11
5.
9
Definition of Diversified Trading & Investment Company
11
Overview and History of Subject Companies
12
Literature Review
15
Portfolio Theory
15
Capital Asset Pricing Model (CAPM)
16
Index Model
17
Accounting Beta
17
Performance Attribution
18
Multi Factor Model
19
Motives of Being a Multi-business Firm
19
Corporate Portfolio Management (CPM)
20
Hypothesis and Methodology
21
3.1.
Scope of Research
21
3.2.
Hypotheses and Framework
21
Hypothesis
21
Framework
22
3.3.
4.
9
Analysis Methodology
23
Portfolio Optimization
23
Index Model
24
Performance Attribution
27
Multi-Factor Model
27
Results
28
4.1.
Organization and Historical Returns of Sumitomo Corporation
28
4.2.
Optimization Result
35
4.3.
Index Model
40
4.4.
Performance Attribution
48
4.5.
Factor Model
51
Interpretation and Discussion
53
5.1.
Analytical Approach
53
5.2.
Comparison between three Possible Portfolio Compositions
54
5.3.
Discussion on Optimal Portfolio Composition
56
6"
"
6.
Comparison between Companies
6.1.
Organization
57
6.2.
History of Diversified Trading and Investment Companies
59
6.3.
Analyses of Subject Companies
61
Mitsubishi Corporation
61
Mitsui & Co., Ltd.
65
Itochu Corporation
69
6.4.
8.
Interpretation and Discussion
73
Mitsubishi Corporation
73
Mitsui & Co
73
Itochu Corporation
73
Optimal Portfolio Composition and History of the Company
77
6.5.
7.
57
Implication for the role of Real Estate
78
Conclusion and Further Discussion
80
Conclusion and Limitations
80
Further Discussions
80
Bibliography
81
7"
"
TABLE OF FIGURES
Figure 1: Historical Net Income of Diversified Trading and Investment Companies"................................................"14"
Figure 2: Historical Market Value of Diversified Trading and Investment Companies"............................................."14"
Figure 3: Organization of Sumitomo Corporation as of July 1, 2011"........................................................................."29"
Figure 4: Historical Quarterly Net Income"................................................................................................................."30"
Figure 5: Historical Annual Net Income"...................................................................................................................."30"
Figure 6: Historical Quarterly ROA"..........................................................................................................................."31"
Figure 7: Historical Annual ROA"..............................................................................................................................."31"
Figure 8: Scatterplots of Quarterly ROAs of each unit"..............................................................................................."33"
Figure 9: Efficient Frontier, Portfolios, and Performance of each unit"......................................................................"36"
Figure 10: Efficient Frontier, Portfolios, and Performance of each unit"....................................................................."38"
Figure 11: Area Chart of Portfolio Weights on the Efficient Frontier (with 2004)"...................................................."39"
Figure 12: Area Chart of Portfolio Weights on the Efficient Frontier (without 2004)"..............................................."39"
Figure 13: Historical Portfolio Weights of Sumitomo Corporation"............................................................................"40"
Figure 14: Historical ROA of Each Segment in the Market"......................................................................................."41"
Figure 15: Historical Market Portfolio Compositions"................................................................................................"42"
Figure 16: Scatterplot of Segment ROA and Market ROA"........................................................................................"45"
Figure 17: Mean Return and Jensen’s Alpha of Each Segment".................................................................................."47"
Figure 18: Systematic Risk-Return Relationship of Each Portfolio"..........................................................................."55"
Figure 19: Organizations of Subject Companies"........................................................................................................"58"
Figure 20: Historical Net Income of Each Segment (Mitsubishi Corporation)".........................................................."62"
Figure 21: Historical ROA of Each Segment (Mitsubishi Corporation)"....................................................................."62"
Figure 22: Efficient Frontier, Portfolios, and Performance of Each Segment (Mitsubishi Corporation)"..................."63"
Figure 23: Area Chart of Portfolio Weights on the Efficient Frontier (Mitsubishi Corporation)"..............................."64"
Figure 24: Historical Net Income of Each Segment (Mitsui & Co)"..........................................................................."66"
Figure 25: Historical ROA of Each Segment (Mitsui & Co)"......................................................................................"66"
Figure 26: Efficient Frontier, Portfolios, and Performance of Each Segment (Mitsui & Co)"...................................."67"
Figure 27: Area Chart of Portfolio Weights on the Efficient Frontier (Mitsui & Co)"................................................"68"
Figure 28: Historical Net Income of Each Segment (Itochu Corporation)"................................................................."70"
Figure 29: Historical ROA of Each Segment (Itochu Corporation)"..........................................................................."70"
Figure 30: Efficient Frontier, Portfolios, and Performance of Each Segment (Itochu Corporation)".........................."71"
Figure 31: Area Chart of Portfolio Weights on the Efficient Frontier (Itochu Corporation)"......................................"72"
Figure 32: Systematic Risk-Return Relationship of Each Portfolio (Mitsubishi)"......................................................."74"
Figure 33: Systematic Risk-Return Relationship of Each Portfolio (Mitsui)"............................................................."75"
Figure 34: Systematic Risk-Return Relationship of Each Portfolio (Itochu)".............................................................."76"
-
-
8"
"
1. Introduction1.1. Research Motivation
Investment. What does one think of when it comes to investment? It could be common
stock, deposit in a bank account, Treasury bill, pension plan, or whatever makes return on initial
invested resources. Everyone has his/her own perception and approach toward investment.
However, no one would disagree that investment is no longer a kind of thing only highly trained
people on Wall Street can handle. Many books for individual investors have been published, and
every kind of information for investment strategy is flooding on the internet. More and more
people think seriously about how to increase their wealth by efficient investment. It is even not
uncommon nowadays that high schools incorporate an investment simulation game in their
curriculum. Investment has already become a part of everyone’s everyday life. Literally, one can
find a new investment or manage his/her portfolio anytime, anywhere. Therefore, something
about investment should be of interest to everyone.
For corporate entities, investment has a more crucial meaning, or technically speaking,
investment is everything. Regardless of its business model, every single corporation needs to
invest its resources in some form of investment opportunity to create outputs, which eventually
bring value to the corporation. Investment could be manufacturing equipment, financial
instruments, developable land, third-party business professional, or again, whatever makes return
and economic sense.
Then all the questions regarding investment finally converge into the following two,
whether individual or institutional: “which opportunities to invest in?” and “how much to invest
in each opportunity?” These foremost questions have been the center of attention for decades, and
extensive research has been made on these topics to date.
9"
"
Based on modern portfolio theory and capital asset pricing model, portfolio compositions
of corporations also have been studied by many researchers to date. When it comes to optimal
portfolio of a diversified firm, the research convention is to assume a hypothetical conglomerate
that invests in various industries in the market, and to use security return indexes of industries as
proxies of the returns of business units of the subject conglomerate. A conglomerate is a firm that
participates in different markets or businesses and grows mainly from acquisition strategy.
Although these studies are full of insights and suggestions, the application is somewhat limited to
diversified firms that do not base their growth on M&A strategy because the security return index
of an industry does not represent the returns of these firms’ business units.
Speaking of returns of diversified firms, a number of studies have examined the
diversification effects by comparison of security or accounting returns of diversified firms to
those of undiversified firms. However, these studies are mostly about the total return of a
diversified firm, and not much has been written on the return of individual business units within a
diversified firm or optimal portfolio composition of the firm. The optimal portfolio of a
diversified firm is more about corporate portfolio management, a relatively new realm of study
which has been developed since the 1970’s initially based more on a strategic management
perspective. More recently, evaluation matrices developed in the corporate portfolio management
field have been synthesized with risk-return measures of portfolio theory, and more
comprehensive studies have been conducted. (Pidun, et al., 2011)
I, as a real estate professional working in one of Japanese diversified trading and
investment companies, have been always curious about how capital should be allocated within the
company. Since the strategy of a business unit should align with the overall strategy of a firm,
understanding how capital can be efficiently allocated to each unit and what role the unit plays in
relationships with other units is of extreme importance.
10"
"
Being motivated by these backgrounds, this paper explores possible optimal corporate
portfolio of diversified firms mainly from financial perspective rather than strategic management
perspective. Although the examples of observation are limited to Japanese diversified trading and
investment firms, the discussions are applicable to virtually any type of corporation. Also, this
paper attempts to examine relationships between the real estate sector and other sectors and to
provide some food for thoughts on how to incorporate real estate sector in a corporate portfolio.
1.2. Approach and Structure
The main questions of this study are: what is the optimal corporate portfolio composition
for a diversified trading and investment Company?, what factors should the management of such
company pay attention to? and which economic or other measures should be used for evaluation
of portfolio?
This paper is divided into 7 chapters. Next section of this chapter presents the definition
and overview of a Japanese diversified trading and investment company, the subject of this study.
Chapter 3 introduces the methodology of the study and Chapter 4 provides the results. Chapter 5
interprets the results, Chapter 6 compares the results of similar companies, and Chapter 7
summarizes the findings.
1.3. Overview-of-Diversified-Trading-and-Investment-CompanyDefinition-of-Diversified-Trading-and-Investment-CompanyJapanese diversified trading & investment companies, the subject of this paper, are
sometimes as to “general trading companies” or “Sogo-Shosha,” and usually distinguished from
“conglomerates.” A common characteristic in both entities is that both participate in multiple
diversified markets or industries. Encyclopedia of Finance defines a conglomerate as “one that
has engaged in several conglomerate combinations” where “a conglomerate combination is a type
of business combination that may involve firms that have little, if any, product market
11"
"
similarities.” (Lee & Lee, 2006) In addition, conglomerate is often characterized to take an
acquisition strategy for its growth.
On the other hand, a distinct characteristic is that the business model of diversified
trading and investment companies were originally trading (e.g. importing raw materials and
exporting finished goods in different industries) and then they gradually integrated their
businesses vertically as well as horizontally. (Lifson, 1981) As its name indicates, they also invest
in both private and public equity and engage in M&A activities as conglomerates do. However,
M&A is not the only source for their growth.
Another characteristic of a diversified trading and investment company is that it is often a
member of a larger conglomerate. In fact, three of the four subject companies are member
companies of larger conglomerate.
There are not so many companies that are qualified as diversified trading and investment
companies defined in this paper, and this paper studies one trading and investment company
(Sumitomo Corporation) in detail and three more companies (Mitsubishi Corporation, Mitsui &
Co, and Itochu Corporation) mainly for comparison.
Overview-and-History-of-Subject-CompaniesSome diversified trading and investment firms have their origin in former-Zaibatsu group,
“any of the large capitalist enterprises of Japan before World War II, similar to cartels or trusts
but usually organized around a single family. One zaibatsu might operate companies in nearly all
important areas of economic activity. The Mitsui combine, for example, owned or had large
investments in companies engaged in banking, foreign trade, mining, insurance, textiles, sugar,
food processing, machinery, and many other fields as well. All zaibatsu owned banks, which they
used as a means for mobilizing capital.” (Encyclopædia Britannica, 2012)"They started importing
and exporting goods, and had gradually evolved to credit enhancement, manufacturing,
12"
"
investment or any other possible form of business. (Lifson, 1981) As the phrase “from noodle to
satellite” represents very well the wideness and the degree of diversification of a diversified
trading and investment company, it is not an exaggeration to say that a diversified trading and
investment company engages in every single industry one can think of.
Figure 1 represents net income of subject companies and it shows that they have grown
rapidly during the past decade. Even after the financial crisis, they have already recovered to the
level before the crisis in terms of net income. In terms of market value, although they have shrunk
almost by a half after the crisis, they maintain their market value at a high level relative to the
market. Figure 2 illustrates the historical market value of the companies, and Table 1 lists the
market value rankings of the subject companies as of July 27, 2012, and it shows that these
companies play significant roles in Japanese Economy.1 2
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
1
"Mitsubishi"Corporation’s"market"value"as"of"date"is"about"USD"31.6"billion,"which"is"equivalent"to"that"of"Lowe’s"
Corporation,"a"home"improvement"retailer."Sumitomo"Corporation’s"market"value"is"about"USD"17"billion,"which"is"
equivalent"to"that"of"Allstate"Corporation,"a"property"and"casualty"insurer."(YCharts,"2012)"
2
"Just"for"reference,"Toyota"Motor"has"the"highest"market"value"of"approximately"USD"126.8"billion,"which"is"
equivalent"to"that"of"Verizon"Communications."(Nikkei"Inc,"2012)"(YCharts,"2012)"
13"
"
Figure 1: Historical Net Income of Diversified Trading and Investment Companies
Figure 2: Historical Market Value of Diversified Trading and Investment Companies
Table 1: Market Value Rankings of Subject Companies
Company"
Ranking"
th
Mitsubishi"Corporation"
14 "
Mitsui"&"Co.,"Ltd."
19 "
Sumitomo"Corporation"
30 "
Itochu"Company"
36 "
th
th
th
"(as"of"July"27,"2012,"source:"Nikkei.com)"
14"
"
2. Literature Review
Portfolio-TheorySince portfolio theory is a widely accepted concept in finance world and a number of
good textbooks and informative articles have been published on this dupject, this paper does not
discuss it in detail. However, it is worthwhile to remember that Markowitz clarified the
relationship between total variance of a portfolio, variance of each security, and the co-movement
between individual securities. (Markowitz, 1952) (Elton & Gruber, 1997) This relationship can be
expressed as:
Equation)1
!
! !! =
!
!! !
!! !, !!!
!!!
!
=
!! !! !"# !! , !!
!!! !!!
Where E(rp) = expected return of a portfolio, E(ri) = expected return of security i, ri = return of
security i, and wi = weight of security i in a portfolio. Since the more securities are incorporated
in a portfolio, the less volatility of each individual security contributes to the total volatility, a
high degree of diversification eliminates variance terms of each security and only covariance
terms remain. The part of total volatility which is diversifiable, is called “idiosyncratic risk”
or ”security specific risk,” and the remaining part, which is not diversifiable is called “systematic
risk” or “market risk.” Since investors can diversify away idiosyncratic risk, only systematic risk
should be rewarded. (Bodie, Kane, & Marcus, 2011) (Brealey, Myers, & Allen, 2011)
Markowitz also formulated how to maximize the expected return of a portfolio for any
given total volatility. He called such a portfolio an “efficient portfolio” and a series of efficient
portfolios as “efficient frontier.” Among efficient portfolios, one that maximizes Sharpe ratio is
called “tangency portfolio” or “market portfolio.” (Sharpe, Mutual Fund Performance, 1966)
(Brealey, Myers, & Allen, Portfolio Theory and the Capital Asset Pricing Model, 2011) The
Sharpe ratio is defined as:
)
)
15"
"
Equation)2
!ℎ!"#$!!"#$% =
! − !!
!"#$!!"#$%&$
=
!"#$%#&%!!"#$%&$'(
!
Capital-Asset-Pricing-Model-(CAPM)Based on Markowitz’ portfolio theory, the capital asset pricing model was developed by
Sharpe, Lintner, and Mossin. (Bodie, Kane, & Marcus, 2011) Their theory is that in a competitive
market, the expected risk premium of a security is proportionate to its market beta, or the
sensitivity to the market. (Sharpe, 1964) (Lintner, 1965) (Mossin, 1966) Under the capital asset
pricing model, the risk-return relationship can be expressed as
Equation)3
! !! − !! = ! ! !! − !! !,
!=
!"# !, !!
!"# !!
where E(ri)=expected+return+of+a+security+i, rf+=risk4free+rate, and E(rm)=expected+return+of+
market+portfolio. Beta is also referred to as a proxy of systematic risk. (Brealey, Myers, & Allen,
2011) From Equation"3, beta of a portfolio is estimated to be following:
Equation)4
!
!! =
!! !!
!!!
In the context of CAPM, there are two major criteria for performance evaluation of a
portfolio: Jensen’s alpha and Treynor ratio. Jensen’s alpha is expressed as:
Equation)5
!! = !! − ! !! = !! − !! + !! !! − !!
and, it is considered to indicate an abnormal return or mispricing of a security. (Jensen, 1967)
fTreynor ratio is similar to Sharpe ratio, but it measures the return of a security to its beta, or
systematic risk instead of its volatility. (Treynor, 1966) (Bodie, Kane, & Marcus, 2011) Treynor
ratio is expressed as
Equation)6
!"#$%&"!!"#$% =
!"#$!!"#$%&$ ! − !!
=
!"#$%&!!"#$
!
16"
"
Index-ModelThe Index model is similar to CAPM, and it is a method to estimate beta of a security.
However, the difference of the two is that index model is a statistical model, and it estimates beta
using a single-variable linear regression, where the independent variable is a market index.
(Bodie, Kane, & Marcus, 2011) The regression equation of index model is:
Equation)7
!! ! − !! = !! + !! !! − !! + !! ! "
where αi=expected+risk+premium+of+a+security+when+market+risk+premium+is+zero, and εi+is+
residual. (Bodie, Kane, & Marcus, 2011) Since εi has zero-mean, risk-return relationship can be
expressed as:
Equation)8)
! !! = !! + !! + !! !! − !! "
Accounting-BetaAccounting beta is the beta estimated with the index model using accounting return
instead of security return. Encyclopedia of Finance expresses accounting beta in the following
equation:
Equation)9)
!"#$
!"#$%!!""#$"
!"#$%&',!,!
= !! + !!!
!"#$
!"#$%!!""#$"
!"#$%&,!
+ !!,! "
Where Aβi=accounting+ beta. This method is used especially when the security is not publicly
traded or when one estimate the sensitivity to the market of a project. (Lee & Lee, 2006)
Theoretically speaking, accounting beta should have a strong correlation with market beta since
in an efficient market all the reported accounting or financial information of firms is reflected in
their stock prices. (Ball & Brown, 1969) Which accounting variable is appropriate for this
analysis has always been a question, and historically, various types of accounting measures such
as profitability, leverage, and liquidity have been adopted. (Bildersee, 1975) "
17"
"
Performance-AttributionPerformance attribution is a performance evaluation technique which measures excess return of a
portfolio in a comparison with the return of a benchmark portfolio called the “Bogey.” This
technique enables investors to identify sources of the excess (or below) market return of their
portfolio. The Bogey can be a market index such as S&P 500, Nikkei 225 or any other portfolio
to which an investor wants to compare his/her own. (Bodie, Kane, & Marcus, 2011) The Bogey
return is calculated as:
Equation)10
!
!! =
!!" !!" ,!
!!!
where wBi+is+weight+of+Sector+i+in+the+Bogey and rBi+is+the+return+of+Sector+i+in+the+Bogey.
The return of the portfolio of an investor is expressed as:
Equation)11)
!
!! =
!!" !!"
!!!
where wPi+is+weight+of+Sector+i+in+the+Portfolio and rPi+is+the+return+of+Sector+i+in+the+Portfolio.
From Equation"10 and Equation"11, the excess return of the portfolio is calculated as:
Equation)12
!
!! − !! =
!
!!" !!" −
!!!
!
!!" !!" =
!!!
(!!" !!" − !!" !!" )
!!!
The total excess return can be decomposed into following three components:
!!" − !!" !!"
Contribution from Sector Allocation
+
Contribution from Project/Investment Selection
+
Contribution from Interaction Effect
=
Total Contribution from Sector i
!!" !!" − !!"
!!" − !!" !!" − !!"
!!" !!" − !!" !!"
They are the potential sources of the excess market return of a portfolio. (Geltner, Miller, Clayton,
& Eichholtz, 2007)
18"
"
MultiOFactor-ModelMulti-Factor model is an analytical method to identify potential risk factors for a security.
In this model, a factor is a surprise, or deviation from its expected value, and macroeconomic
indicator which seems to have an impact on a specific security is often used as a factor. Since
each factor is a surprise, if there is no surprise in any factor, the realized return of a security
should be equal to its expected return. Although there is a debate how to forecast the expected
return, the CAPM expected return is often used as an input. (Bodie, Kane, & Marcus, 2011)
(Chan, Karceski, & Lodonishok, 1998) The Multi-Factor model can be summarized as:
Equation)13)
!! = ! !! + !!! !! + !!! !! + ⋯ + !!" !! + !! "
Motives-of-Being-a-MultiObusiness-FirmNumerous researches have been conducted on conglomerate activity and diversification
effects to date. In the first place, motives of engaging in conglomerate activity can be categorized
into three main realms: profitability, synergism, and diversification. (Smith & Schreiner, 1969)
The profitability motive suggests that a conglomerate firm should enter into a new
industry if the expected return of any investment opportunity in that industry exceeds the cost of
capital of the firm. (Smith & Schreiner, 1969)
The synergism motive explains higher expected return due to economies of scale realized
either from cost reduction by efficient management and operation or demand increase by
combining businesses. The average return of conglomerate firms operating in related businesses
tends to outperform the average return of firms operating in unrelated businesses. (Bettis & Hall,
1982)
The diversification motive emphasizes the reduction in total volatility of corporate
portfolio realized by operating businesses in different industry categories. (Smith & Schreiner,
19"
"
1969) On the contrary, Conglomerate firms have higher market risk than comparable nonconglomerate firms. (Melicher & Rush, 1973) However, related diversification tends to lead to
lower systematic risk than unrelated diversification. (Lubatkin & Rogers, 1989)
Corporate-Portfolio-Management-(CPM)Since its invention in the late 1960s, corporate portfolio management has been discussed
and developed by a number of scholars and practitioners. In the early years, strategic consulting
firms led to invent several risk matrices represented by Boston Consulting Group’s growth-share
matrix and GE/Mckinsey nine-block matrix, and CPM was aimed at evaluating a specific market
for a firm by analyzing market attractiveness and relative positioning of the firm in the industry.
These matrices helped firms make decisions regarding scope of business, capital allocation within
portfolio, and overall firm strategy. (Henderson, 1973) (Pidun, et al., 2011) (Untiedt & Pidun,
2011)
In more recent years, CPM has been enhanced to incorporate risk-return measures, and
the focus of CPM shifted from evaluation of performance and strategy of each business unit to
risk-return management of overall portfolio, which eventually affects the strategy of a firm.
(Pidun, et al., 2011) (Untiedt & Pidun, 2011) A synthesis of strategic management theories and
financial portfolio theories is the key to further develop CPM.
20"
"
3. Hypothesis and Methodology
3.1. Scope of Research
From data availability and economic impact perspectives, the subject of this research is
limited to public diversified trading and investment companies in Japan. Specifically, this paper
studies following four companies:
• Sumitomo Corporation
• Mitsubishi Corporation
• Mitsui & Co, and
• Itochu Corporation
Sumitomo Corporation is the main subject. Of the subject companies, Sumitomo,
Mitsubishi and Mitsui are member companies of Japanese conglomerates, and Itochu is
independent of any. In addition, these companies have all distinct histories and origins, and this
combination enables a well-balanced comparison.
3.2. Hypotheses and Framework
HypothesisIn management and finance fields, it is well established that the primary goal of a
corporation is to maximize shareholders’ wealth, or in other words, the equity value of the firm.
Stakeholder theory suggests expanding the scope of this goal to the stakeholders, which include
shareholders, employees, government, customers, suppliers, and more. While there is a debate
which shareholders or stakeholders a firm should be managed for, different countries have
different attitudes toward this question. (Brealey, Myers, & Allen, 2011) Taking into account the
Japanese social and business norms, this paper consider a firm should maximize stakeholders’
wealth. Above all stakeholders, the shareholders and employees are the most important for a firm.
In terms of shareholders’ wealth maximization, a firm tries to maximize its equity value.
Since equity value is a function of market value, in order to maximize its market value, a firm
21"
"
tries to lower its discount rate or cost of capital. Because individual stock investors can diversify
their portfolio and eliminate idiosyncratic risk of each security, idiosyncratic risk is not rewarded
and the discount rate of a firm is determined based on its systematic risk. Holding a firm’s return
or income constant, lower systematic risk leads to lower discount rate and hence higher market
value. (Bodie, Kane, & Marcus, 2011) (Brealey, Myers, & Allen, 2011) Also, excess market
return generates positive NPV and increases the firm’s value. Therefore a firm is motivated to
keep its systematic risk low relative to its return and to invest more capital in business units with
positive excess market return.
As to employees’ value maximization, a firm aims fundamentally to stabilize its
management and operations to maintain its employment level stable and to protect its employees
from the risk of layoffs. In order to stabilize its management and operations, a firm tries to
maintain not only its systematic risk but also its total risk, which is measured by volatility of
return. From stability perspective, a firm wants to avoid having its return very sensitive to
specific risk factor. In addition, since higher compensation increases employees’ wealth, a firm
seeks excess market return in order to produce additional compensation for its employees.3 4
FrameworkBased on the hypotheses discussed in the previous section, this paper evaluates
performances of corporate portfolios by using following measures and further discusses possible
approaches to compose the optimal portfolio.
Measures: Sharpe ratio, Accounting Beta, Jensen’s Alpha, Treynor Ratio, Excess Market Return,
and Macroeconomic Factor Beta
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
3
"A" firm" can" pay" higher" compensation" for" its" employees" if" it" earns" the" same" return" with" fewer" employees," or" in"
other"words"if"it"raises"the"labor"efficiency."However,"this"approach"is"more"of"an"organizational"matter,"and"this"
paper"does"not"focus"on"the"matter."
4
"Other" than" these" aspects," social" recognition" and" reputation" of" a" firm" can" possibly" affects" its" firm" value" for"
employees.""
22"
"
3.3. Analysis Methodology
Portfolio Optimization
Since this paper deals with corporate portfolio and business units within a company
which does not issue any security, accounting return is used as a proxy of each unit return. All
subject companies publish their earnings (net income) and total assets by business unit in their
quarterly earnings announcement. This paper defines the accounting return used for analysis to be
net income over total assets, which is referred to as ROA hereafter. The analysis period will be 11
years from fiscal year 2001 to fiscal year 2011 due to the data availability of Sumitomo
Corporation, the main subject of this paper.
Based on the mean, standard deviation, and correlations of historical ROAs of each
business unit, Markowitz’s portfolio optimization is conducted for each company with a “no
short-sell constraint” since it is unrealistic to assume that a corporation can short-sell one or more
of its business units. In portfolio optimization process, efficient frontier, tangency portfolio, and
“Suboptimal Portfolio,” a possible portfolio which this paper defines as a portfolio with the same
volatility as the current portfolio and on the efficient frontier, are identified.5 Also, portfolio
compositions on the efficient frontier are shown in an area chart to visualize relationship between
portfolio composition and target return of overall portfolio.
Because of organizational restructuring, subject companies do not maintain exactly the
same business units all through the analysis period. Historical organizational charts of subject
companies illustrates that these companies tend to merge smaller business units into a larger one.
Therefore, both net income and total assets of smaller units before restructuring are summed up
into one based on the current organization for analysis purpose.
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
5
"Because"investors"accept"current"level"of"return"and"volatility"and"investors"can"diversify"the"idiosyncratic"risk"of"a"
security"away,"it"makes"more"sense"to"have"higher"return"with"the"same"volatility"than"to"have"lower"return"with"
lower"volatility.""
23"
"
Since two of the subject companies have a few business units based on geographical
region not on product or industry, net income and total assets of such units are allocated to the
other units proportionate to their net income and total assets under the assumption that these
geographical units also have the same types of business and structure.6 7
For optimization, both quarterly and annual data are collected. Quarterly data provides
more data points and is better for analysis purpose. In spite of its higher frequency, accounting
return of some businesses tends to have some seasonality and can overestimate the volatility of
the business. Therefore, this paper adopts either quarterly or annual data with the higher average
correlation (covariance) for optimization because more highly correlated data gives a more
conservative and risk averse result, which is better especially under uncertain circumstances.
Index Model
The systematic risk of each business unit is estimated by index model using ROA.
Accounting beta of each segment is obtained by linear regression where the dependent variable is
ROA of each segment and the independent variable is ROA of the market index which will be
defined later.
Equation 14
!! ! − !! (!) = !! + !! !! ! − !! (!) + !! (!)
The market index is replicated by synthesizing the historical ROAs of constituent
companies of Nikkei 225, a Japanese market representing stock index. First, historical constituent
companies are identified by the data from Compustat, and each company is classified into
categories which Nikkei defines by the nature of its businesses. Each category is classified into
industries, and each industry to segments that correspond to the organization of Sumitomo
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
6
"Sumitomo"Corporation"has"“Domestic"Regional"Business"Units"and"Offices”"and"“Overseas"Subsidiaries"and"
Branches.”"Mitsui"&"Co"has"“Americas,”"“Europe,"the"Middle"East"and"Africa,”"and"“Asia"Pacific.”"
7
"This"paper"excludes"“Others”"or"“Corporate"Adjustment”"from"analysis."
24"
"
Corporation, as listed in Table 2. Net income and total assets of each constituent company are
collected via Datastream, and ROAs of each category are defined as follows:
Equation)15
!"#!!"!!"#ℎ!!"#$%& =
!"#!!"#$%&!
!!
!"#$%!!""#$"!
where Net+Incomei=Net+Income+of+Company+i+in+the+Sector, and Total+Assetsi=Total+Assets+
of+Company+i+in+the+Sector.
Next, the ROA of each industry is calculated based upon the market value weighted
average of each category as shown below;
Equation)16
!"#!!"!!"#$%&'( =
!"#! ×
!"!
,
!"!
where ROAi=ROA+of+Sector+i+in+the+industry, and MVi=Aggregate+Market+Value+of+Sectori+in+
the+industry. ROA of segment and the market respectively is calculated likewise.8
As to the frequency of market index return, since only annual data is available on
Datastream, annual ROA is used as inputs for this analysis.
After estimating accounting beta of each business unit, Jensen’s Alpha, and Treynor
Ratio are also estimated to evaluate the performance of the unit.
Although this is not the main focus of this paper, a comparison between betas of subject
companies and the market averages is made in order to examine the diversification effect that is
expected to decrease the systematic risk of a diversified company.
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
8
"ROA"of"each"segment"is"used"as"an"input"for"performance"attribution"analysis."
25"
"
Table 2: Index Model Segments, Industries, and Categories
Corresponding+Segments
Industries
Metal+Products
Transportation
Infrastructure+&+Machinerty
Media+&+Lifestyle+Retail
Nonferrous+Metals
Metal
Steel+Products
Automotive
Automotive
Ship+Building
Ship+Building
Machinery
Machinery
Media
Communications
Lifestyle+&+Retail
Chemicals
Textile+&+Apparel
Electric+Machinery
Precision+Instruments
Electric+Power
Energy
Gas
Mineral+Resources
Life+Science
Construction+&+Real+Estate
Foods,+Materials+&+Real+Estate
Retail
Chemicals
Electronics
Mineral+Resources,+Energy,
Chemicals+&+Electronics
Nikkei+225+Categories
Mining
Oil&Coal+Products
Pharmaceuticals
Construction
Real+Estate
Fishery
Foods
Foods
Glass&Ceramics
Materials
Pulp+&+Paper
Rubber+Products
Air+Transport
Marine+Transport
Logistics
Other+Land+Transport
Warehousing
Logistics+&+Financial+Services
Banking
Financial+Services
Insurance
Other+Financial+Services
Securities
Others
Services
Services
Other+Mnufacturing
Other+Mnufacturing
Railway/Bus
Railway/Bus
Trading
Trading+Companies
26"
"
Performance Attribution
Based on the historical mean return of each business unit and the Bogey portfolio, excess
market return of each unit is calculated by Equation 12 in page 18. The Bogey is constructed
based on the corresponding segments listed in Table 2.
Since this paper aims to consider optimal corporate portfolios looking forward, the
analysis uses the current portfolio composition of subject companies and the market rather than
the historical average.
Multi-Factor Model
In addition to the Single Index Model, Multi-Factor Model analysis is conducted in order
to identify possible risk factors to each segment. Considering the diversified and global nature of
subject companies’ businesses, the broad-brush macroeconomic factors and foreign currency
exchange rates are adopted as potential risk factors: Inflation rate, GDP Growth rate, USD/JPY
exchange rate, EUR/JPY exchange rate, CNY/JPY exchange rate, 1-year Japanese Government
bond yield, and 10-year Japanese Government bond yield.
Factor betas of each segment are obtained from multivariate regression as shown below:
Equation)17
!! − ! !! = !! + !!"#$ !!"# + !!"#$ !!"# + !!"#$ !!"# + !!"#$ !!"# + !!"#$ !!"# + !!!! !!! + !!!"! !!"!
where αi=intercept+coefficient, βi+n=factor+beta+of+Segment+i+to+Factor+n,
FINF=%+change+in+inflation+rate,
FGDP=%+change+in+GDP+growth+rate,
FUSD/JPY=%+change+in+USD/JPY+exchange+rate,++
FEUR/JPY=%+change+in+EUR/JPY+exchange+rate,++
FCNY/JPY=%+change+in+CNY/JPY+exchange+rate,++
F1Y=%+change+in+14year+Japanese+Government+bond+yield, and++
F10Y=%+change+in+104year+Japanese+Government+bond+yield.
27"
"
4. Results
4.1. Organization and Historical Returns of Sumitomo Corporation
The organizational chart of Sumitomo Corporation is presented in Figure 3, and the
optimal portfolio weights composition between these seven business units is the main focus of
this paper. Figure 4 and Figure 5 respectively illustrates quarterly and annual historical net
income of each unit. A large part of total net income comes from the Mineral unit, followed by
Transportation and Media. Figure 6 and Figure 7 shows quarterly ROA and annual ROA, and
they are also summarized in Table 3 and Table 4. Figure 8 demonstrates scatterplots of quarterly
ROA.
As Figure 6, Figure 7 and Figure 8 indicate, the Media unit and General Products and
Real Estate unit both seem to have outlying performance in 2004, especially in 4th quarter. This is
because they “recognized gain on issuance of stock by Jupiter Telecommunications, (their)
associated companies listed on the Jasdaq securities exchange, and impairment loss on real estate
for rent in Yokohama area. In addition, equity in earnings of associated companies increased by
16.7 billion yen to 37.4 billion yen mainly contributed by the strong performances of Batu hijau
copper and gold mine project and Jupiter Telecommunications.” (Sumitomo Corporation, 2005)
The company may have restated the real estate asset because it had enough surplus
from stock issuance to offset the loss incurred by the restatement. However, it is difficult to judge
whether this outlying data is the result of arbitrary managerial decision or if it represents a true
characteristic of the business. Such kind of managerial decision could possibly affect the mean
return and volatility of a business unit and also correlations between units, and could lead to
misinterpretation of true nature of a business. This eventually could alter the optimal portfolio
composition. However, it should be noted that this kind of characteristics of accounting return
may be considered as a part of the risk of a company. If so, then taking out such outliers means
underestimating the true volatility of the company or its businesses.
28"
"
Figure 3: Organization of Sumitomo Corporation as of July 1, 2011
(Sumitomo"Corporation,"2011)"
29"
"
Figure 4: Historical Quarterly Net Income
Figure 5: Historical Annual Net Income
30"
"
Figure 6: Historical Quarterly ROA
Figure 7: Historical Annual ROA
31"
"
Table 3: Summary Statistics of Quarterly ROA
Metal
Trans
Infra
Media
Mineral
GeneRE
NewInd
TTL
N
mean
sd
Min
max
.0070973
.0045326
.0047543
.0066787
.0093464
.0045025
.0029657
.0055443
44
.0040751
.0028067
.0042932
.0087831
.0074938
.0043083
.0030442
.003117
-.00303
-.002923
-.0061831
-.0104779
-.0049444
-.0112138
-.0044215
-.0017141
.0175585
.0106102
.0127461
.0503589
.0315792
.0181431
.0085462
.0112228
mean
sd
Min
max
.0276523
.0182309
.0183938
.0262284
.0368327
.0172586
.0113841
.0217296
11
.0112792
.008194
.0143765
.0149811
.0210021
.0092617
.0082457
.0095585
.0145139
.0065421
-.0002045
.0095755
.0096235
-.0029414
-.0034395
.0058145
.0503874
.0323112
.042044
.0662528
.0750872
.0282191
.0261417
.0346855
Table 4: Summary Statistics of Annual ROA
Metal
Trans
Infra
Media
Mineral
GeneRE
NewInd
TTL
N
This matter should be carefully treated because the decision to include this kind of
outlying data gives different results especially in terms of portfolio optimization.
From volatility perspective, such outliers should be included if they represent the true
characteristics of the businesses since higher volatility gives more conservative decisions which
are good in terms of risk management. However, especially when the number of data points is
limited, those outliers affect the result to a great degree, so this kind of question should be
carefully handled. This paper solves for Markowitz portfolio optimization both with and without
4Q2004 data, and compares the results.
As the data inputs for portfolio optimization, this study uses annual data because of its
higher average correlation than quarterly data as seen in Table 5 and Table 6. Annual data also has
the higher average volatility relative to its time horizon, and this seems adequate from risk
aversion perspective.
32"
"
Figure 8: Scatterplots of Quarterly ROAs of each unit
33"
"
Table 5: Correlation Matrix of Quarterly ROA
Metal
Trans
Infra
Media
Mineral
GeneRE
NewInd
Average
Rank
Average
Metal
1
0.6566
0.2210
90.0778
0.3338
0.3031
90.0180
0.2364
6
0.1686
Trans
0.6566
1
0.3236
90.0684
0.4756
0.4795
0.1979
0.3441
7
Infra
0.2210
0.3236
1
0.0110
0.1438
0.2725
90.2000
0.1286
3
Media
90.0778
90.0684
0.0110
1
0.0584
90.4147
0.3701
90.0202
1
Mineral
0.3338
0.4756
0.1438
0.0584
1
0.2673
0.1262
0.2342
5
GeneRE
0.3031
0.4795
0.2725
90.4147
0.2673
1
0.0783
0.1643
4
NewInd
90.0180
0.1979
90.2000
0.3701
0.1262
0.0783
1
0.0924
2
Table 6: Correlation Matrix of Annual ROA
Metal
Trans
Infra
Media
Mineral
GeneRE
NewInd
0.6193
0.7258
0.0869
0.2661
0.2629
=0.1835
Metal
1
Trans
0.6193
1
0.6282
0.1923
0.7444
0.5692
0.4665
Infra
0.7258
0.6282
1
=0.1545
0.4045
0.6284
=0.2708
Media
0.0869
0.1923
=0.1545
1
0.1431
=0.4675
0.2766
Mineral
0.4045
0.1431
1
0.4140
0.2818
0.2661
0.7444
GeneRE
0.2629
0.5692
0.6284
=0.4675
0.4140
1
0.1362
NewInd
=0.1835
0.4665
=0.2708
0.2766
0.2818
0.1362
1
Average
0.2963
0.5366
0.3269
0.0128
0.3757
0.2572
0.1178
4
7
5
1
6
3
2
Rank
Average
0.2748
The summary statistics and the average correlation of Annual ROA excluding 2004
data are listed respectively in Table 7 and Table 8. The average correlation becomes
approximately 13 points higher as expected due to the elimination of outlying performance of
Media unit and General Products & Real Estate unit. On the contrary, in spite of the elimination,
total volatility is higher without 2004 data. In terms of corporate risk management, data without
2004 seems better since it has both higher volatility and higher average correlation. However, this
paper analyzes both data as stated earlier because of the sensitivity of whether or not to identify
2004 data as an outlier.
34"
"
Table 7: Summary Statistics of Annual ROA without 2004
Metal
Trans
Infra
Media
Mineral
GeneRE
NewInd
TTL
N
mean
sd
min
max
.0272959
.0183348
.0193021
.022226
.0374221
.0192786
.0114189
.0223651
.0118239
.0086296
.0148177
.0073196
.022042
.0067408
.0086909
.0098275
.0145139
.0065421
-.0002045
.0095755
.0096235
.0101948
-.0034395
.0058145
.0503874
.0323112
.042044
.0331932
.0750872
.0282191
.0261417
.0346855
10
Table 8: Correlation Matrix of Annual ROA without 2004 data
Metal
Trans
Infra
Media
Mineral
GeneRE
NewInd
0.6277
0.7690
:0.0129
0.2786
0.4933
:0.1830
Metal
1
Trans
0.6277
1
0.6340
0.4956
0.7444
0.7810
0.4663
Infra
0.7690
0.6340
1
0.0688
0.3955
0.7063
:0.2799
:0.0129
0.4956
0.0688
1
0.4887
0.5419
0.6236
0.3955
0.4887
1
0.5043
0.2817
Media
Mineral
0.2786
0.7444
GeneRE
0.4933
0.7810
0.7063
0.5419
0.5043
1
0.1826
NewInd
:0.1830
0.4663
:0.2799
0.6236
0.2817
0.1826
1
Average
0.3288
0.6248
0.3823
0.3676
0.4489
0.5349
0.1819
2
7
4
3
5
6
1
Rank
Average
0.4099
4.2. Optimization Result
Figure 9 depicts the efficient frontier, current portfolio, tangency portfolio, suboptimal
portfolio, and risk-return relationships of business units as a result of portfolio optimization with
2004. These three portfolios are compared in Table 9.
Tangency Portfolio consists mainly of Metal (22.04%), Media (26.82%) and General
Products & Real Estate (41.72%). In addition to high returns of Metal (2.76%) and Media
(2.62%), these three have very low correlations to one another and hence very low covariance.
suboptimal portfolio consists mainly of Metal (50.62%), Media (21.62%) and Mineral (21.62%).
35"
"
Figure 9: Efficient Frontier, Portfolios, and Performance of each unit
Table 9: Comparison between three portfolios (with 2004 data)
WeightsJ(%)
Expected Standard
Return Deviation
(%)
(%)
Sharpe
Ratio
Metal
Trans.
Infra.
Media
Mineral
Gene.JRE
NewJInd.
Current
11.35
16.01
10.01
18.33
20.82
13.71
9.77
2.173
0.956
2.273
Tangency
22.04
0.00
0.00
26.82
0.00
41.72
9.42
2.140
0.565
3.785
Suboptimal
50.62
0.00
0.00
25.63
21.62
2.13
0.00
2.905
0.955
3.042
Suboptimal Portfolio has higher weight in Mineral instead of General Products & Real Estate in
order to increase the portfolio return. However, regardless of its highest volatility of all,
correlations between Mineral and the other two are very low (Metal: 0.27, Media: 0.14), and that
works to maintain the volatility still at a level as low as the current portfolio (0.96%). As to
General Products & Real Estate, it works very well mainly to diversify the total volatility away
with low correlations with the other high-performing units.
36"
"
Figure 10 depicts the efficient frontier, current portfolio, tangency portfolio, Suboptimal
Portfolio, and risk-return relationship of each business unit as a result of portfolio optimization
without 2004. Table 10 compares current, tangency, and suboptimal portfolios.
Taking out the 2004 data gives different results especially for Media and General
Products & Real Estate since it changes the risk profile of the two and correlations with the other
units as well. Although General Products & Real Estate has higher ROA than with 2004, it now
has high correlations with high-performing units such as Mineral (0.50), Metal (0.49), and Media
(0.54), so there is no room for the unit to play the diversifier role any longer. This is reflected in
its 0% weight both in tangency and suboptimal portfolios.
On the other hand, Media still has low average correlation (0.37) and high mean
return (2.22%) at the same time, and it plays a significant role especially for tangency Portfolio as
a diversifier. Without 2004, it seems that Media completely replaces General Products & Real
Estate’s position with 2004. However, Media now has high correlation with Mineral (0.49), and
this is the main reason why it has lower weight especially when the target portfolio return reaches
the upper range as its weight in suboptimal portfolio illustrates.
It is worthwhile to mention that there is no notable difference between the suboptimal
portfolios of both cases. Both portfolios consist mainly of Metal (with 2004: 50.62%, without
2004: 46.55%), Media (25.63%, 27.23%), and Mineral (21.62%, 26.22%). Figure 11 and Figure 12
represents the portfolio compositions on the efficient frontier for any given target ROA for
analyses both with and without 2004, and they illustrate how handling data 2004 alters portfolio
compositions. Historical portfolio composition is also presented in Figure 13 for comparison
purpose.
37"
"
Figure 10: Efficient Frontier, Portfolios, and Performance of each unit
Table 10: Comparison between three portfolios (without 2004)
WeightsJ(%)
Expected Standard
Return Deviation
(%)
(%)
Sharpe
Ratio
Metal
Trans.
Infra.
Media
Mineral
Gene.JRE
NewJInd.
Current
11.35
16.01
10.01
18.33
20.82
13.71
9.77
2.237
0.983
2.276
Tangency
30.53
0.00
0.00
69.47
0.00
0.00
0.00
2.377
0.620
3.836
Suboptimal
46.55
0.00
0.00
27.23
26.22
0.00
0.00
2.857
0.982
2.910
Since this paper bases Suboptimal Portfolio and discuss how one can improve it, whether
data with or without 2004 makes only a slight difference, so further analysis is conducted based
only on the data without 2004, which provides more conservative and risk averse insight because
of its higher average correlation.
38"
"
Figure 11: Area Chart of Portfolio Weights on the Efficient Frontier (with 2004)
Figure 12: Area Chart of Portfolio Weights on the Efficient Frontier (without 2004)
39"
"
Figure 13: Historical Portfolio Weights of Sumitomo Corporation
4.3. Index Model
Summary statistics of ROA of each market segment and the entire market sorted
according to Table 2 is presented in Table 11. Lines in Figure 14 represent historical ROA of
market segments. The Sharpe ratio of each segment is also calculated in Table 12. As the tables
and the figure indicate, Media segment has the highest mean return (3.13%), and New Industry
Development segment has the lowest (0.34%) mainly because of the low return of Banking
category. As to Sharpe ratio, Others is the highest (2.32), and Metal Products is the lowest (0.73).
Figure 15 shows historical market portfolio compositions calculated based on market
value by segment are shown as an area chart. As seen from the figure, Mineral segment has by far
the largest share in the market.
40"
"
Table 11: Summary Statistics of Historical ROA of Each Segment in the Market
mean
sd
min
max
Metal
.0213522
.0292968
-.0128496
.0644852
Trans
.0296839
.0192367
-.0128703
.0469208
Infra
.019726
.0152368
-.0044452
.0424473
Media
.0312662
.0182504
-.0178555
.0516605
Mineral
.0186322
.0159079
-.0122309
.0368617
GeneRE
.0192327
.0095744
.0017532
.0285201
NewInd
.003445
.0043425
-.0026287
.0094695
Others
.0226551
.009759
.0043931
.0358048
Market
.0197238
.0117829
.000604
.0339283
N
11
Figure 14: Historical ROA of Each Segment in the Market
41"
"
Table 12: Sharpe Ratio of Each Segment in the Market
(1)
(2)
(3)=(1)/(2)
Historical
Standard
Sharpe
Historical
Sharpe
Mean
Deviation
Ratio
Mean
Ratio
Metal;Products
0.02135
0.02930
0.72882
4
8
Transportation;&;Construction;
Systems
0.02968
0.01924
1.54309
2
4
Infrastructure
0.01973
0.01524
1.29463
5
5
Media,;Network;&;Lifestyle;Retail
0.03127
0.01825
1.71318
1
3
Mineral;Resources,;Energy,;Chemical;
&;Electronics
0.01863
0.01591
1.17126
7
6
General;Products;&;Real;Estate
0.01923
0.00957
2.00876
6
2
New;Industry;Development;&;CrossU
function
0.00344
0.00434
0.79331
8
7
Others
0.02266
0.00976
2.32145
3
1
Figure 15: Historical Market Portfolio Compositions
42"
"
Ranking
Risk premiums of segments without 2004 data are summarized in Table 13. Table 14 lists
Ordinary Least-square Regression (OLS) results. With constant, none of the beta coefficients are
statistically significant, and two of the constant terms are not statistically significant either.
Another alternative is Regression Through the Origin (RTO), the regression analysis which
suppresses constant term. On one hand, Index Model has a constant term by definition as shown
in Equation 14. On the other hand, RTO seems more appropriate approach if it gives less standard
error and better fit to the model than OLS does. (Hahn, 1977) (Eisenhauer, 2003)
Then RTO is also conducted and Table 15 shows the results. Figure 16 illustrates the
relationships between ROA of each segment and the market ROA as a scatterplot, and the slope
of each fitted line represents the accounting beta of each segment, which measures the systematic
risk of each. As seen in Table 15, all the segments have statistically significant beta coefficients,
and they all have less standard error and much higher R-square than OLS results. Therefore, this
paper concludes RTO is appropriate, and further analyzes the data based on the RTO results.
Nevertheless, this matter should be carefully handled and may require further discussion since
whether OLS or RTO is used for analysis could alter the conclusion of this paper. However,
because of the availability of historical returns of the subject and the market, the number of
sample data is very limited, and no more precision can be expected.
Table 16 presents accounting beta of each market segment to the market, and the
comparison to Table 15 indicates that Sumitomo Corporation has lower accounting beta than
industry average in five segments (Metal, Transportation, Infrastructure, Media, and General
Products & Real Estate), and higher accounting betas in the other two segments (Mineral and
New Industry Development).9
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
9
"This" is" not" the" main" topic" of" this" paper," but" the" results" seems" to" be" inconsistent" with" the" theory" that"
diversification"increases"systematic"risk"of"diversifying"company,"and"further"study"may"be"of"interest.""
43"
"
Table 13: Summary Statistics of Historical Risk Premium of Each Segment of Sumitomo Corporation
Metal
Trans
Infra
Media
Mineral
GeneRE
NewInd
TTL
N
mean
sd
min
max
.0249703
.0160091
.0169765
.0199004
.0350965
.016953
.0090933
.0200395
.0097972
.0074139
.0125265
.0079124
.0215015
.0057964
.0095754
.0085575
.0127534
.0061638
-.0003997
.0082402
.0092451
.0099995
-.008835
.0056193
.0449919
.031058
.0353845
.0325116
.073834
.0270123
.0248886
.0334324
10
Table 14: Accounting Betas with Constant
Market
_cons
(1)
Metal
(2)
Trans
(3)
Infra
(4)
Media
(5)
Mineral
(6)
GeneRE
(7)
NewInd
(8)
TTL
-0.03121
(0.9189)
0.02549**
(0.0026)
0.2112
(0.3486)
0.01251*
(0.0180)
0.02403
(0.9511)
0.01658
(0.0596)
0.3473
(0.1308)
0.01416**
(0.0087)
-0.05130
(0.9392)
0.03595*
(0.0243)
0.2672
(0.1093)
0.01253**
(0.0028)
0.3912
(0.1642)
0.002622
(0.6199)
0.1246
(0.6394)
0.01798**
(0.0077)
10
0.0005
10
0.2616
10
0.0008
10
0.2886
10
0.2267
10
0.0288
N
10
10
2
R
0.0014
0.1102
p-values in parentheses
*
p < 0.05, ** p < 0.01, *** p < 0.001
Table 15: Accounting Betas without Constant
Market
(1)
Metal
(2)
Trans
(3)
Infra
(4)
Media
(5)
Mineral
(6)
GeneRE
(7)
NewInd
(8)
TTL
1.0338**
(0.0054)
0.7342**
(0.0014)
0.7168*
(0.0192)
0.9388***
(0.0004)
1.4508*
(0.0142)
0.7910***
(0.0003)
0.5008**
(0.0051)
0.8759**
(0.0027)
10
0.4737
10
0.7712
10
0.5056
10
0.7796
10
0.6009
10
0.6496
(4)
Media
(5)
Mineral
(6)
GeneRE
(7)
NewInd
(8)
Others
N
10
10
2
R
0.5959
0.6979
p-values in parentheses
*
p < 0.05, ** p < 0.01, *** p < 0.001
Table 16: Accounting Betas of Market Segments
(1)
(2)
(3)
Metal
Trans
Infra
Market
N
R2
1.3165***
(0.0005)
1.4109***
(0.0000)
1.0059***
(0.0000)
1.4691***
(0.0000)
1.0317***
(0.0000)
0.9091***
(0.0000)
0.2193***
(0.0000)
1.0249***
(0.0000)
11
0.7224
11
0.8424
11
0.8687
11
0.8685
11
0.9501
11
0.9395
11
0.8544
11
0.9023
p-values in parentheses
*
p < 0.05, ** p < 0.01, *** p < 0.001
44"
"
Figure 16: Scatterplot of Segment ROA and Market ROA
45"
"
The Accounting beta, Jensen’s Alpha, and the Treynor ratio of each segment are
summarized in Table 17 and Table 18. Figure 17 depicts those measures graphically. 10 As to
rankings, smaller number indicates higher mean return, Sharpe ratio, Jensen’s Alpha, and Treynor
ratio, and lower beta. Mineral has the highest accounting beta (1.45), and New Industry
Development has the lowest (0.50). Top three high accounting beta segments (Mineral, Metal,
and Media) are also the top three high mean return segments, and there seems to be a strong
correlation between mean return and accounting beta, which means higher returns are due to
higher systematic risk. This can be examined by following Jensen’s Alpha and Treynor ratio
analysis.
As to Jensen’s Alpha, it is noteworthy that all segments have positive alphas. Above all,
Mineral is the highest (0.93%) followed by Metal (0.76%) and Infrastructure (0.48%), and it
indicates these segments outperform the market the most in absolute term. As to Treynor ratio,
Metal is the highest (0.027) followed by Infrastructure and Mineral, and it indicates these
segments outperform the market the most relative to their systematic risks.
Large alphas and high Treynor ratios in Metal, Media, and Mineral unit indicate that high
performing units have high systematic risk but also high alpha both in relative and absolute terms.
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
10
"Index" model" analysis" was" also" made" based" on" estimated" EBIT" of" each" segment" and" Market" EBIT" to" exclude"
financial"leverage"and"income"tax"effect."While"all"segments"got"lower"betas"with"EBIT"than"with"net"income,"which"
is" consistent" with" the" theory," relative" rankings" were" unchanged." While" net" income" of" each" segment" is" officially"
announced" by" the" company," EBIT" is" estimated" based" on" net" income" and" total" assets" of" each" segment" and" net"
interest" expenses" and" income" taxes" of" the" entire" company," this" paper" focuses" on" net" income" approach" for"
precision."
46"
"
Table 17: Accounting Beta of Each Segment
(1)
Historical
Mean
(2)
Standard
Deviation
(3)=(1)/(2)
Sharpe
Ratio
(4)
Accounting
Beta
(1)
Mean
Ranking
(3)
Sh.
(4)
Beta
MetalAProducts
0.02730
0.01182
2.30854
1.03384
2
3
6
TransportationA&AConstructionA
Systems
0.01833
0.00863
2.12464
0.73423
6
4
3
Infrastructure
0.01930
0.01482
1.30264
0.71685
4
7
2
Media,ANetworkA&ALifestyleARetail
0.02223
0.00732
3.03650
0.93885
3
1
5
MineralAResources,AEnergy,A
ChemicalA&AElectronics
0.03742
0.02204
1.69776
1.45080
1
5
7
GeneralAProductsA&ARealAEstate
0.01928
0.00674
2.85999
0.79095
5
2
4
NewAIndustryADevelopmentA&A
CrossWfunction
0.01142
0.00869
1.31389
0.50080
7
6
1
Table 18: Jensen’s Alpha and Treynor Ratio of Each Segment
Ranking
(1)
Historical
Mean
(5)
CAPM
Forecast
(6)=(1)'(5)
Jensen's
Alpha
(7)=(1)/(4)
Treynor
Ratio
Jensen's
Alpha
Treynor
Ratio
MetalBProducts
0.02730
0.02035
0.00694
0.02640
2
2
TransportationB&BConstructionB
Systems
0.01833
0.01479
0.00354
0.02497
5
4
Infrastructure
0.01930
0.01447
0.00483
0.02693
3
1
Media,BNetworkB&BLifestyleBRetail
0.02223
0.01859
0.00364
0.02367
4
6
MineralBResources,BEnergy,B
ChemicalB&BElectronics
0.03742
0.02809
0.00933
0.02579
1
3
GeneralBProductsB&BRealBEstate
0.01928
0.01584
0.00344
0.02437
6
5
NewBIndustryBDevelopmentB&B
Cross'function
0.01142
0.01046
0.00096
0.02280
7
7
Figure 17: Mean Return and Jensen’s Alpha of Each Segment
47"
"
4.4. Performance Attribution
The results of performance attribution are presented in Table 19, Table 20, Table 21, and
Table 22. As seen in Table 19, a large part of the contribution of segment allocation comes from
Metal and Media. As a whole, it seems that the company has negative active weight in poorly
performing segment such as New Industry, and allocates more capital to high-performing
segment such as Media, which leads to the overall positive contribution of segment allocation.
As seen in Table 20, Mineral has by far the highest excess market return (1.73%).
Mineral also has the highest market weight, and this is why it contributes a great deal to the
overall excess market return which comes from project/investment selection. New Industry and
Metal have the next highest excess market return, but because of its low market weight, the
contribution of Metal is limited. On the other hand, Transportation and Media have negative
excess market return, and they are the least contributing segments.
As to interaction effect, Mineral makes huge and negative contribution (-0.13%) because
it has lower weight than market regardless of its large excess market return. So does New
Industry (-0.04%). On the contrary, Media contributes negatively (-0.05%) because it has higher
weight than market although it has negative excess market return as well as Transportation (0.03%).
As seen in Table 22, in aggregate, Metal and Mineral lead the total excess market return
respectively by 0.24% and 0.21%, followed by Infrastructure (0.11%), and Transportation and
Media have negative contributions of -0.10% and -0.01%. General Products & Real Estate has a
slightly positive contribution of 0.05%.
48"
"
Table 19: Contribution of Segment Allocation
(1)
(2)
SegmentBAllocation
WeightsB(%)
Segment
Portfolio
Market
(3)=(1)'(2)
Active
Weights
(%)
(4)
Index
Performance
(%)
MetalBProduct
11.35
3.39
7.95
2.1352
0.1698
114.99%
Transportation
16.01
13.22
2.79
2.9684
0.0828
56.05%
Infrastructure
10.01
4.09
5.92
1.9726
0.1168
79.08%
Media
18.33
13.32
5.01
3.1266
0.1567
106.07%
MineralBResources
20.82
28.16
'7.35
2.0163
'0.1481
'100.28%
GeneralBProductsB&BRealB
Estate
13.71
11.40
2.31
1.9233
0.0444
30.07%
9.77
15.09
'5.32
0.3445
'0.0183
'12.42%
'
11.31
'11.31
2.2655
'0.2563
'173.55%
0.1477
100.00%
NewBIndustry
Others
BBBBBBBBB
ContributionBofBSegmentBAllocation
Table 20: Contribution of Project/Investment Selection
(1)
(2)
(3)=(1)'(2)
Portfolio
Index
Excess
Performance Performance Performance
Segment
(%)
(%)
(%)
(4)
Market
Weight
(%)
(5)=(3)x(4)
Project
Selection
Contribution
MetalAProduct
2.7296
2.1352
0.5944
3.39
0.0202
20.46%
Transportation
1.8335
2.9684
'1.1349
13.22
'0.1500
'152.26%
Infrastructure
1.9302
1.9726
'0.0424
4.09
'0.0017
'1.76%
Media
2.2226
3.1266
'0.9040
13.32
'0.1205
'122.23%
MineralAResources
3.7422
2.0163
1.7259
28.16
0.4861
493.25%
GeneralAProductsA&ARealA
Estate
1.9279
1.9233
0.0046
11.40
0.0005
0.53%
NewAIndustry
1.1419
0.3445
0.7974
15.09
0.1204
122.12%
2.2655
'2.2655
11.31
'0.2563
'260.12%
0.0985
100.00%
Others
AAAAAAAAAAA
'
ContributionAofAIndividualAProjectsAwithinASectors
49"
"
(5)=(3)x(4)
Segment
Allocation
Contribution
Table 21: Contribution of Interaction Effect
(3)
Active
Weights
(%)
Segment
(7)
Excess
Performance
(%)
(9)=(3)x(7)
Interaction
Effect
Contribution
Metal<Product
7.95
0.5944
0.0473
85.92%
Transportation
2.79
H1.1349
H0.0317
H57.52%
Infrastructure
5.92
H0.0424
H0.0025
H4.56%
Media
5.01
H0.9040
H0.0453
H82.32%
H7.35
1.7259
H0.1268
H230.42%
2.31
0.0046
0.0001
0.19%
H5.32
0.7974
H0.0424
H77.14%
H11.31
H2.2655
0.2563
465.85%
0.0550
100.00%
Mineral<Resources
General<Products<&<Real<Estate
New<Industry
Others
Contribution<of<Interaction<Effect
Table 22: Total Contribution
Segment
(5)
(8)
(9)
Segment
Project
Interaction
Allocation
Selection
Effect
Contribution Contribution Contribution
Metal<Product
0.1698
0.0202
0.0473
0.2373
78.76%
Transportation
0.0828
J0.1500
J0.0317
J0.0989
J32.83%
Infrastructure
0.1168
J0.0017
J0.0025
0.1126
37.36%
Media
0.1567
J0.1205
J0.0453
J0.0091
J3.01%
J0.1481
0.4861
J0.1268
0.2112
70.09%
0.0444
0.0005
0.0001
0.0450
14.95%
New<Industry
J0.0183
0.1204
J0.0424
0.0596
19.77%
Others
J0.2563
J0.2563
0.2563
J0.2563
J85.08%
0.3013
100.00%
Mineral<Resources
General<Products<&<Real<Estate
Total<Contribution
50"
"
(10)=
(5)+(8)+(9)
Total
Contribution
4.5. Multi-Factor Model
Annual factors are summarized in Table 23, and estimates of beta coefficients of each
segment to risk factors are listed in Table 24. While Infrastructure and Mineral Resources
segments have several statistically significant betas, none of the potential factors has consistently
significant beta coefficients with all the segments. Because of the statistical insignificance of
these results, these factors are not used for further analyses.
This finding may be partially explained by the nature of accounting return. Some of
businesses have seasonality, and some of the factors may have a lagged influence in accounting
return. Therefore, in order to conclude the relationship between these potential risk factors and
accounting return of each segment, further investigation is necessary. However, in addition to the
fact that the number of observation is very limited, the return of each segment is the aggregate of
several different businesses because of the organization. Hence it is difficult to clarify the
relationship especially in this study, and this paper does not discuss the matter any further. 11
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
11
"Multicfactor" model" analysis" was" also" conducted" with" 2004" data" or" based" or" percentage" changes" in" difference"
between" realized" ROA" and" expected" ROA." However" results" were" more" or" less" similar" to" ones" presented" in" this"
section"and"not"statistically"significant"in"either"case."
51"
"
Table 23: Summary Statistics of Annual Factors
Inflation
GDP
USDJPY
EURJPY
CNYJPY
Short
Long
mean
sd
min
max
-1.675999
3.830072
-12.58183
2.387527
12.62454
44.59932
-7.439031
146.825
-.0388846
.0648428
-.1575057
.0534835
.0020481
.0853313
-.1367273
.0957651
-.0179482
.0535549
-.0725586
.0726033
.6985265
2.167079
-.6737025
6.851877
-.0109838
.1650036
-.1857227
.3566692
N
11
Table 24: Beta Coefficient of Each Segment to Risk Factors (Annual)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
Metal
Trans
Infra
Media
Mineral
GeneRE
NewInd
TTL
0.002587
0.001618
0.002681*
-0.000129
0.000941
0.000819
0.001096
0.001507
(0.1959)
(0.1336)
(0.0417)
(0.8846)
(0.0832)
(0.5392)
(0.3682)
(0.1292)
-0.000048
0.000099
-0.000043
0.000048
0.00022**
0.000029
0.000133
0.000068
(0.7010)
(0.1845)
(0.5066)
(0.4763)
(0.0037)
(0.7575)
(0.1761)
(0.2846)
-0.08580
-0.06353
-0.2371*
0.04136
-0.1479*
-0.1288
0.07330
-0.09493
(0.5000)
(0.3456)
(0.0244)
(0.5312)
(0.0113)
(0.2278)
(0.3978)
(0.1660)
-0.1097
-0.08390
-0.1110*
-0.09650
-0.2176***
-0.05388
-0.03932
-0.1168*
(0.2051)
(0.0932)
(0.0474)
(0.0728)
(0.0009)
(0.3735)
(0.4482)
(0.0343)
0.2415
0.1330
0.3988*
0.03648
0.09570
0.1647
-0.04503
0.1663
(0.2668)
(0.2360)
(0.0207)
(0.7202)
(0.1060)
(0.3090)
(0.7280)
(0.1369)
-0.000911
0.000164
0.000728
-0.004857
0.003271*
0.001496
-0.000090
0.000356
(0.8291)
(0.9388)
(0.7330)
(0.0959)
(0.0372)
(0.6462)
(0.9743)
(0.8558)
-0.003455
-0.01109
-0.01593
0.08460*
-0.01911
-0.05260
-0.01515
-0.02026
(0.9320)
(0.5989)
(0.4568)
(0.0225)
(0.1177)
(0.1601)
(0.5840)
(0.3265)
0.01675
0.001917
0.003214
0.01447
0.01189*
-0.002974
-0.004103
0.004793
(0.1595)
(0.7026)
(0.5276)
(0.0541)
(0.0112)
(0.6926)
(0.5416)
(0.3346)
N
11
11
11
11
11
11
11
11
R2
0.6854
0.8621
0.9416
0.8977
0.9932
0.7745
0.7111
0.9175
Inflation
GDP
USDJPY
EURJPY
CNYJPY
Short
Long
_cons
p-values in parentheses
*
p < 0.05, ** p < 0.01, *** p < 0.001
52"
"
5. Interpretation and Discussion
5.1. Analytical Approach
This chapter develops some thoughts on possible optimal corporate portfolio based on the
three portfolios identified in the previous chapters: current portfolio, tangency portfolio, and
suboptimal portfolio.
It is worthwhile to remember that the goal of private enterprise is to maximize the value
for stakeholders: shareholders and employees. For shareholders, firm value is the most important
indicator, and for employees stability of management or their employment and higher
compensation are important factors.
From the firm value maximization perspective, lower systematic risk is desirable since
only systematic risk should be compensated in capital markets, and the lower systematic risk
leads to the lower discount rate to be applied to the company. Also, the higher excess market
return, the higher firm value since the excess brings about positive NPV for shareholders.
These two factors are also very crucial for employees as well. Lower systematic risk
means more stability of management or operations of the company, and it enables more stable
employment. On the other hand, higher excess market return enhances the ability of the company
to compensate its employees more.
Another important aspect is the degree of diversification and the combination of
businesses since diversification is thought to have an effect on company to decrease its total risk
and to increase its return especially when diversified businesses are closely related.
"
-
53"
"
5.2. Comparison-between-three-Possible-Portfolio-CompositionsTable 25 lists the portfolio compositions of current, tangency, and suboptimal portfolios.
Table 26 summarizes risk-return measures including expected return, expected volatility, and
other ratios for each portfolio. These portfolios are also graphically illustrated in Figure 18.
In a comparison with current portfolio, tangency portfolio does not make much sense
because investors can improve the Sharpe ratio of their own portfolio by combining Sumitomo
Corporation’s stock and other stocks or investments and diversifying away the idiosyncratic risk.
Rather, systematic risk, or accounting beta matters since investors cannot decrease the systematic
risk of the company they invest in by themselves. Holding systematic risk constant, investors
would always choose higher return investment opportunities. Conversely, investors have an
incentive to invest in a stock with the same volatility but with the lower systematic risk, which
leads to higher Treynor ratio.
In this respect, although its Sharpe ratio is not the highest, suboptimal portfolio has the
higher Treynor ratio than tangency portfolio does, and it has still a higher Sharpe ratio than the
current portfolio does, so it makes more economic sense to shareholders and also to employees.
In addition, the suboptimal portfolio has by far the highest Jensen’s alpha and also excess
market return estimated based on performance attribution approach, which is the source of
positive NPV for shareholders and better welfare for employees.
From the perspectives stated above, suboptimal portfolio seems a very good composition.
However, it weighs heavily on three segments: Metal (46.55%), Media (27.23%), and Mineral
(26.22%), and seems rather unbalanced in terms of diversification. Since this paper assumes that
ex-post return of each unit, which is an input for optimization, counts diversification effects, such
as reduction in total risk and excess market return, unbalanced portfolios are less likely to achieve
the ex-ante return estimated in this framework.
54"
"
Table 25: Comparison between three Portfolio Compositions (Sumitomo Corporation)
Segment
Current
Tangency
Subopti mal
Portfol i o
Portfol i o
Portfol i o
Wei ghts
Wei ghts
Wei ghts
(%)
(%)
(%)
Metal <Product
11.35
30.53
46.55
Transportati on
16.01
0.00
0.00
Infrastructure
10.01
0.00
0.00
Medi a
18.33
69.47
27.23
Mi neral <Resources
20.82
0.00
26.22
General <Products<&<Real <Estate
13.71
0.00
0.00
9.77
0.00
0.00
New<Industry
Table 26: Comparison between Indicators of three Portfolios (Sumitomo Corporation)12
Current
Tangency
Subopti mal
Portfol i o
Portfol i o
Portfol i o
Wei ghts
Wei ghts
Wei ghts
(%)
(%)
(%)
ExpectedEReturnEbasedEonEHi stori cal EMean
2.2365
2.3774
2.8570
StandardEDevi ati on
0.9828
0.6198
0.9816
SharpeERati o
2.2758
3.8358
2.9104
Accounti ngEBeta
0.9381
0.9678
1.1173
CAPMEforecastedEReturn
1.8576
1.9127
2.1901
Jensen'sEAl pha
0.3789
0.4647
0.6669
TreynorERati o
2.3827
2.4552
2.5561
Total EExcessEMarketEReturn
0.3013
0.3575
0.7994
Figure 18: Systematic Risk-Return Relationship of Each Portfolio
"
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
12
"Accounting" beta" of" current" portfolio" is" different" presented" in" this" table" is" weighted" sum" and" different" from"
accounting"beta"of"total"return"obtained"by"index"model."
55"
"
5.3. Discussion-on-Optimal-Portfolio-CompositionSince Metal, Media, and Mineral all have higher betas to the market, the other four
divisions are favorable in terms of systematic risk management. From an excess market return
perspective, Mineral, Metal and Media have the three highest alphas, and these results support the
portfolio optimization results. According to the performance attribution results, Mineral has by
far the highest excess market return which is 1.73%, followed by New Industry (0.80%) and
Metal (0.59%). Media has negative excess market return (-0.90%) and so does Transportation (1.13%), so these two division should be discounted in terms of portfolio weights.
Synthesizing these results, Mineral and Metal should be highly weighted. The correlation
between these two is low, and this is also a good combination in terms of the portfolio total
volatility. Because of its negative excess market return and relatively high accounting beta, Media
should be less weighted than suboptimal portfolio suggests, and its weight should be replaced by
segments with low beta and yet positive excess market return such as New Industry Development
(beta: 0.50, excess return: 0.80%) and General Products and Real Estate (0.79, 0.005%).
Also it should be noted that all the data inputs are ex-post mean and volatility, and that
these values are not necessarily good estimates for ex-ante forecast. With that noted, since
Metal’s weight in suboptimal (46.55%) is extremely high and the performance of the portfolio is
largely affected by the precision Metal’s forecast, Metal should be less weighted than suggested.
In addition to these three perspectives, the weights of the other units should be
determined and the total portfolio composition should be adjusted with the consideration of
diversification effects and the degree of relatedness between business units. 13
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
13
"Since"each"segment"consists"of"multiple"subcsegments"operating"in"different"industries,"it"is"difficult"to"conclude"
the"degree"of"relatedness"between"segments."Also,"it"is"challenging"to"quantify"how"much"diversification"improves"
systematic" risk" and" realized" return," so" this" paper" does" not" further" discuss" how" to" incorporate" this" matter" into"
optimal"corporate"portfolio"composition."
56"
"
7. Comparison-between-Companies6.1. OrganizationFigure 19 illustrates the organization of each company and the product lines that the
business units of each company deal in.14 Although all the companies cover most of the same
products, they all have slightly different organization from one another, which makes it difficult
to conduct an exact apple-to-apple comparison in terms of unit return. However, their
organizations are more or less similar, and they are comparable enough to grasp the optimal
corporate portfolio of each company.
Another consideration is that none of the companies has a business unit that engages
solely in real estate. For instance, General Products & Real Estate segment of Sumitomo
Corporation deals in real estate together with food, fertilizer, and other general products such as
construction materials. Mitsubishi Corporation has a segment called Industrial Finance, Logistics
& Development, and the segment’s businesses are financial services, real estate, and logistics.
Furthermore, the combination of businesses is not necessarily related and or rather unrelated and
diversified in some cases in order to decrease the total volatility of the unit, which makes it even
more difficult to analyze the effect of having real estate in a portfolio. Therefore, this paper
discusses the role of real estate in a portfolio in a limited manner.
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
14
"Because"of"its"organizational"restructuring"in"2011,"Itochu"Corporation"currently"has"separate"two"segments"for"
Construction" &" Realty," and" Financial" Service" and" Insurance" Services" &" Logistics." This" paper" handles" these" two"
segments"jointly"to"have"more"data"points"for"analysis"purpose.""
57"
"
Figure 19: Organizations of Subject Companies
Products
Sumitomo
Mitsubishi
Mitsui
Itochu
Metal>Products
Metals
Iron &>Steel>Products
Energy,>Metals
&>Minerals
Machinery
Machinery>&
Infrastructure>Projects
ICT>&>Machinery
Metal
Steel
Transportation
Construction
Systems
Transportation
&
Construction Systems
Machinery
Infrastructure
Infrastructure
Textile
Media
Textile
Media, Network
&
Lifestyle>Retail
Living>Essentials
Foods>&>Retail
Retail
Energy
Energy
Mineral
Consumer>Service &>
IT
Mineral Resources,
Energy,>Chemicals
&>Electronics
Energy
Mineral>&
Metal>Resources
Chemicals
Chemicals
Chemicals
Food
General
Products
Food
General Products
&
Real>Estate
&>General
Merchandise
Real>Estate
Financial
Services
Logistics
Chemicals,
Forest Products
New> Industry>
Development
&
CrossHfunction
Industrial Finance,
Logistics
&>Development
Logistics &
Financial>Business
Construction,>Realty,
Financial>and
Insurance
& Logistics
(Sumitomo"Corporation,"2011)"(Mitsubishi"Corporation,"2011)"(Mitsui"&"Co,"2011)"(Itochu"Corporation,"2011)"
58"
"
6.2. History-of-Diversified-Trading-&-Investment-CompaniesHistories of four subject companies are briefly summarized in Table 27. As one can see
in the organizations of subject companies in Figure 19, they have more or less similar businesses
today, and there is no notable difference between them except for profitability of each segment
and corporate portfolio compositions.
Therefore, this paper focuses on the earlier history of each company before they came to
have similar businesses, especially on the origin or the primary business at the beginning of the
company. Moreover, since three of the four are members of larger conglomerate, or former
Zaibatsu, and the characteristics of the conglomerate they belong to possibly affects businesses or
organization of the company because of their close business relationships with other member
companies, this paper also sheds lights on the history of such organization especially before the
World War II.
Sumitomo Group has its origin in copper mining in smelting business. Another
characteristic is that Sumitomo Corporation started its operation as a real estate company, and
that it made a transit to general trading company right after the WWII. (Sumitomo Corporation,
2012)
Mitsubishi Group originated in shipping business of general merchandise, and it was the
first Japanese company to open an overseas commerce route. In 1880’s, it expanded its businesses
into diversified industries in Meiji Era. (mitsubishi.com committee, 2012) (Mitsubishi
Corporation, 2012)
Mitsui Group started as textile shops, and it has its origin in retail business. Mitsui & Co
is said to be the first Japanese general trading company, and its primary business was trading in
rice and coal at its origin. (Encyclopædia Britannica, 2012) (Mitsui & Co., Ltd., 2012)
59"
"
Table 27: Histories of Subject Companies
time
before
1700
Sumitomo
Mitsubishi
Mitsui
Book and medicine
shop is founded as
the origin of
Sumitomo
Itochu
Textile shops are
founded as the origin
of Mitsui
Expanded into
copper mining &
smelting
from
1850
to
1900
from
1900
to
1945
1946
Expanded into
machinery, mining,
manufacturing, and
banking
Former Sumitomo
Corporation is
founded as a real
estate company
(1919)
Makes a transition to
general trading
(1945)
A shipping firm is
founded as the origin
of Mitsubishi (1870)
Expanded into
banking, trading, and
mining
Expanded into
mining, shipbuilding,
banking, insurance
and warehousing
Former Mitsui & Co
is founded trading in
rice & coal (1876)
Expanded into paper
& glass
manufacturing,
brewery and heavy
industries such as
machinery, electrical
equipment,
chemicals, and
automobile
Founded as a linen
trading company
(1858)
Expanded to heavy
industries including
machinery, chemicals
and mining
“Zaibatsu” is dissolved
Expanded into
general trading
1970’s
Itochu is the only one of the four that does not belong to any conglomerate. It started its
business in linen trading and specialized in textile trading business for a long time until it
expanded its operations into broader industries in 1970s. (Itochu Corporation, 2012)
60"
"
6.3. Analyses-of-Subject-CompaniesMitsubishi Corporation
Figure 20, Figure 21, Table 28, and Table 29 respectively shows historical net income,
historical ROA, summary statistics of ROA, and correlations of each segment ROA of Mitsubishi
Corporation. Figure 22 depicts the efficient frontier, current portfolio, tangency portfolio,
Suboptimal Portfolio, and risk-return relationship of each business unit as a result of portfolio
optimization. Figure 23 represents portfolio compositions on the efficient frontier for any given
target ROA, and Table 30 summarizes current, tangency, and suboptimal portfolio compositions.
Finally, the accounting beta of each segment as a result of index model analysis is presented in
Table 31.
In spite of its high return (4.31%) and large contribution to net income in absolute term,
the Metals segment is excluded as a result of portfolio optimization mainly because of its high
volatility (2.28%) and high correlation with Energy segment (0.84) which has the highest return
and plays a significant role for suboptimal portfolio (77.44%). The other segment which is
incorporated in suboptimal portfolio is Living Essentials segment (22.56%). Although it has
relatively low return (2.23%), it has the lowest average correlation (0.03) and negative correlation
with Energy (-0.26), and it works well to diversify the idiosyncratic risk away.
As to accounting beta, Energy is the highest (1.94) followed by Metals (1.70) and
Chemicals (1.23). On the contrary, Industry Finance, Logistics and Developments has the lowest
accounting beta (0.55) followed by Living Essentials (0.90) and Machinery (0.99).
61"
"
Figure 20: Historical Net Income of Each Segment (Mitsubishi Corporation)
Figure 21: Historical ROA of Each Segment (Mitsubishi Corporation)
62"
"
Table 28: Summary Statistics of Historical ROA (Mitsubishi Corporation)
Indust
Energy
Metals
Machi
Chemi
Living
TTL
N
mean
sd
min
max
.0041695
.0498548
.0430738
.0211386
.0316911
.022341
.026756
.0222358
.0165383
.0228432
.0101293
.0118702
.00308
.0126376
-.049247
.0260232
.0114407
.0062542
.0118133
.0154141
.0073941
.0280677
.0756765
.0763854
.0331926
.0459987
.0260695
.0412093
11
Table 29: Correlation Matrix of Annual ROA (Mitsubishi Corporation)
Indust.
Energy
Metals
Machi.
Chemi.
Living.
1
0.1078
;0.0407
0.8330
;0.0697
0.7543
Indust.
Energy
0.1078
1
0.8421
0.5244
0.9001
;0.2623
Metals
;0.0407
0.8421
1
0.4345
0.7275
;0.5019
Machi.
0.8330
0.5244
0.4345
1
0.3234
0.4797
Chemi.
;0.0697
0.9001
0.7275
0.3234
1
;0.3050
Living.
0.7543
;0.2623
;0.5019
0.4797
;0.3050
1
Average
0.2641
0.3520
0.2436
0.4325
0.2627
0.0275
4
5
2
6
3
1
Rank
Average
0.2637
Figure 22: Efficient Frontier, Portfolios, and Performance of Each Segment (Mitsubishi Corporation)
63"
"
Figure 23: Area Chart of Portfolio Weights on the Efficient Frontier (Mitsubishi Corporation)
Table 30: Comparison between three portfolios (Mitsubishi Corporation)
WeightsJ(%)
Expected
Return
(%)
Standard
Deviation
(%)
Sharpe
Ratio
Indust.
Energy
Metals
Machi.
Chemi.
Living.
Current
7.78
14.29
32.01
17.33
7.23
21.36
2.676
1.264
2.117
Tangency
0.00
0.00
7.12
0.00
2.96
89.92
2.409
0.248
9.712
Suboptimal
0.00
77.44
0.00
0.00
0.00
22.56
4.365
1.264
3.452
Table 31: Accounting Beta of Each Segment (Mitsubishi Corporation)
Market
N
R2
(1)
Indust
(2)
Energy
(3)
Metals
(4)
Machi
(5)
Chemi
(6)
Living
(7)
TTL
0.5455
1.9423***
1.7007**
0.9855***
1.2272***
0.8990***
1.1342***
(0.0514)
(0.0006)
(0.0018)
(0.0000)
(0.0009)
(0.0001)
(0.0002)
11
11
11
11
11
11
11
0.6397
0.9264
0.6852
0.8201
0.7698
0.3284
0.7109
p-values in parentheses
*
p < 0.05, ** p < 0.01, *** p < 0.001
64"
"
Mitsui & Co., Ltd.
Figure 24, Figure 25, Table 32, and Table 33 respectively show historical net income,
historical ROA, summary statistics of ROA, and correlations of each segment ROA of Mitsui &
Co. Figure 26 depicts the efficient frontier, current portfolio, tangency portfolio, Suboptimal
Portfolio, and risk-return relationship of each business unit as a result of portfolio optimization.
Figure 27 represents portfolio compositions on the efficient frontier for any given target ROA,
and Table 34 summarizes current, tangency, and suboptimal portfolio compositions. Finally, the
accounting beta of each segment as a result of index model analysis is presented in Table 35.
In terms of net income, it is clear that the company is driven largely by Mineral and
Energy. While both Mineral and Energy have high ROA (12.85%, 6.47%), they have very high
volatility (5.08%, 3.17%). Figure 26 illustrates these units’ risk-return relationships. Because of
their high volatility, Mineral and Energy are moderately weighted for suboptimal portfolio since
suboptimal portfolio has as low volatility as 1.13%, which is far lower than those of the two units.
On the other hand, despite of its relatively low ROA (2.04%), Machinery is weighted as much as
50.54% mainly because of its second lowest average correlation (0.11) and its low or negative
correlations with Mineral (0.12) and Energy (-0.73).
As to systematic risk, Mineral has by far the highest accounting beta (4.73), and Energy
is the second highest (2.03). On the other hand, Foods has an extremely low beta (0.02), and
Consumer and Logistics and Finance also have very low betas (0.19, 0.39), and systematic risk of
the company seems to come largely from Mineral and Energy.
65"
"
Figure 24: Historical Net Income of Each Segment (Mitsui & Co)
Figure 25: Historical ROA of Each Segment (Mitsui & Co)
66"
"
Table 32: Summary Statistics of Historical ROA (Mitsui & Co)
Iron
Mineral
Machi
Chemical
Energy
Foods
Consum
LogiFin
TTL
N
mean
sd
Min
max
.0173012
.1285041
.0203877
.0124157
.0646965
.0016887
-.012037
.0037748
.0289431
7
.0192062
.0507819
.0075727
.015373
.0317173
.0134257
.03938
.0173106
.0102941
-.0199279
.064875
.0135203
-.021769
.0317131
-.0197608
-.0734354
-.0285345
.015775
.0352706
.1879126
.0348911
.0212157
.1129669
.0213036
.0218568
.022621
.0435451
Table 33: Correlation Matrix of Annual ROA (Mitsui & Co)
Iron
Iron
1
Mineral
Machi.
Chemical
Energy
Foods
Consum.
0.0358
0.4099
0.8453
?0.5244
?0.0631
0.8200
Logi.7Fin.
0.9363
Mineral
0.0358
1
0.1203
0.0460
0.4811
0.7298
?0.2363
?0.0631
Machi.
0.4099
0.1203
1
0.3884
?0.7336
?0.1503
0.6415
0.1344
Chemical
0.8453
0.0460
0.3884
1
?0.5779
0.0018
0.7570
0.7868
Energy
?0.5244
0.4811
?0.7336
?0.5779
1
0.5842
?0.8727
?0.3630
Foods
?0.0631
0.7298
?0.1503
0.0018
0.5842
1
?0.3981
?0.1404
Consum.
0.8200
?0.2363
0.6415
0.7570
?0.8727
?0.3981
1
0.6569
Logi.7Fin.
0.9363
?0.0631
0.1344
0.7868
?0.3630
?0.1404
0.6569
1
Average
0.4100
0.1961
0.1127
0.2434
?0.2739
0.1174
0.1186
0.2152
8
5
2
7
1
3
4
Rank
Average
0.1424
Figure 26: Efficient Frontier, Portfolios, and Performance of Each Segment (Mitsui & Co)
67"
"
6
Figure 27: Area Chart of Portfolio Weights on the Efficient Frontier (Mitsui & Co)
Table 34: Comparison between three Portfolios (Mitsui & Co)
WeightsH(%)
Expected Standard
Sharpe
Return Deviation
Consum. Logi.HFin.
Ratio
(%)
(%)
Iron
Mineral
Machi.
Chemical
Energy
Foods
Current
7.24
15.50
18.53
9.48
24.19
10.55
8.91
5.59
3.139
1.130
2.777
Tangency
0.00
0.00
64.97
6.53
22.85
0.00
5.64
0.00
2.816
0.388
7.251
Suboptimal
0.00
13.08
50.54
0.00
29.05
0.00
7.33
0.00
4.502
1.113
4.044
Table 35: Accounting Beta of Each Segment (Mitsui & Co)
(1)
(2)
Iron
Market
N
2
0.9577
(3)
Mineral
***
4.7296
**
(4)
Machi
0.8107
**
(5)
(6)
(7)
(8)
(9)
Energy
Foods
Consum
LogiFin
TTL
2.0295
0.02149
0.1933
0.3942
1.1331**
Chemical
0.6664
**
(0.0007)
(0.0096)
(0.0017)
(0.0069)
(0.0574)
(0.9223)
(0.7737)
(0.1321)
(0.0023)
7
7
7
7
7
7
7
7
7
0.7297
0.4782
0.0017
0.0149
0.3360
0.8104
R
0.8734
0.7002
0.8288
p-values in parentheses
*
p < 0.05, ** p < 0.01, *** p < 0.001
68"
"
Itochu Corporation
Figure 28, Figure 29, Table 36, and Table 37 respectively show historical net income,
historical ROA, summary statistics of ROA, correlations of each segment ROA of Itochu
Corporation. Figure 30 depicts the efficient frontier, current portfolio, tangency portfolio,
Suboptimal Portfolio, and risk-return relationship of each business unit as a result of portfolio
optimization. Figure 31 represents portfolio compositions on the efficient frontier for any given
target ROA, and Table 38 summarizes current, tangency, and suboptimal portfolio compositions.
Finally, accounting beta of each segment as a result of index model analysis is presented in Table
39.
As Figure 28 illustrates, a large part of net income comes from Energy segment. In terms
of ROA, Energy is also the highest performing segment (7.08%), but at the same time the most
volatile (3.52%) as is usual the case. While Textile has the second highest mean ROA (4.25%), it
has a lower volatility (1.34%) than the entire portfolio (1.65%), and hence a high Sharpe ratio.
Suboptimal portfolio consists mainly of Textile segment (78.50%), and Energy is
incorporated into it (20.06%) to increase the portfolio return. Since Energy has the highest
volatility and very high correlation with Textile (0.76), Machinery is slightly incorporated
(1.44%) in order to diversify away the idiosyncratic risk. In fact, Machinery has the lowest
average correlation (-0.017) and very low or negative correlations with Textile (-0.30) and
Energy (0.016).
As for accounting beta, Energy is the highest (2.96), followed by Textile (1.63) and
Chemicals (1.12). On the other side of spectrum, Construction, Finance and Logistics segment
has a negative beta (-0.69), and Food has the second lowest beta (0.63).
69"
"
Figure 28: Historical Net Income of Each Segment (Itochu Corporation)
Figure 29: Historical ROA of Each Segment (Itochu Corporation)
70"
"
Table 36: Summary Statistics of Historical ROA (Itochu Corporation)
Textile
Machi
Energy
Chemi
Food
ConstRE
TTL
N
mean
sd
min
Max
.042484
.0204362
.0707633
.0264908
.0176217
-.0165012
.0233049
11
.0133612
.0128408
.0351712
.010415
.0113991
.0431455
.0164714
.0216271
-.0065738
.0237085
-.0003979
-.0127452
-.1240395
-.0071188
.0635295
.0337621
.1153386
.0352918
.0337487
.0257502
.0461799
Table 37: Correlation Matrix of Annual ROA (Itochu Corporation)
Textile
ICT)Machi.
Energy
Chemi.
Food
Cost.)RE
1
;0.3011
0.7595
0.6118
0.2867
0.4267
;0.3011
1
0.0158
;0.1369
;0.0767
0.3979
Energy
0.7595
0.0158
1
0.6460
0.2471
0.3799
Chemi.
0.6118
;0.1369
0.6460
1
;0.0021
0.1582
Food
0.2867
;0.0767
0.2471
;0.0021
1
0.0072
Cost.)RE
0.4267
0.3979
0.3799
0.1582
0.0072
1
Average
0.2973
;0.0169
0.3414
0.2128
0.0770
0.2283
5
1
6
3
2
4
Textile
ICT)Machi.
Rank
Average
0.1900
Figure 30: Efficient Frontier, Portfolios, and Performance of Each Segment (Itochu Corporation)
71"
"
Figure 31: Area Chart of Portfolio Weights on the Efficient Frontier (Itochu Corporation)
Table 38: Comparison between three Portfolios (Itochu Corporation)
WeightsJ(%)
Standard
Deviation
(%)
Sharpe
Ratio
Textile
Machi.
Energy
Chemi.
Food
7.19
19.57
30.48
16.24
21.56
4.96
2.330
1.647
1.415
Tangency
39.82
34.69
0.00
14.31
11.17
0.00
2.977
0.697
4.269
Suboptimal
78.50
1.44
20.06
0.00
0.00
0.00
4.784
1.647
2.905
Current
Cost.JRE
Expected
Return
(%)
Table 39: Accounting Beta of Each Segment (Itochu Corporation)
Market
N
R2
(1)
Textile
(2)
Machi
(3)
Energy
(4)
Chemi
(5)
Food
(6)
ConstRE
(7)
TTL
1.6291***
0.8936***
2.9551***
1.1228***
0.6283*
-0.6846
0.9780**
(0.0008)
(0.0004)
(0.0004)
(0.0001)
(0.0132)
(0.2639)
(0.0022)
11
11
11
11
11
11
11
0.7338
0.8116
0.4745
0.1229
0.6241
0.6951
0.7249
p-values in parentheses
*
p < 0.05, ** p < 0.01, *** p < 0.001
72"
"
6.4. Interpretation-and-Discussion-A comparison between current, tangency, and suboptimal portfolios is also conducted for
these three companies, and the results are presented in Table 42 to Table 45. 15 16 Also, systematic
risk-return relationships or the companies are graphically illustrated in Figure 32 to Figure 34.
Mitsubishi-CorporationThe high accounting beta of the suboptimal portfolio comes mainly from Energy (1.94)
although the other component, Living Essentials has the second lowest (0.8990) of all its
segments. In order to mitigate its high systematic risk, low beta segments such as Industrial
Finance (0.55) or Machinery (0.99) seems appropriate to be incorporated.
Mitsui-&-CoDespite the very high betas of Mineral (4.73) and Energy (2.03), the relatively low beta
of Machinery (0.81) offsets the total systematic risk to some degree. However, other segments
such as Foods, Consumer Service and Logistics & Finance have very low betas (0.02, 0.19, 0.39),
and they are the possible candidates to have more weight in the optimal portfolio.
Itochu-CorporationThe two major components, Textile and Energy both have very high betas (1.63, 2.96),
and the weight of Machinery with a relatively low beta (0.89) seems too low to mitigate the
systematic risk of the entire portfolio. From an accounting beta perspective, Construction &
Realty and Financial & Logistics has a negative beta (-0.68) and has a high potential to decrease
the systematic risk. However, it is the only segment with a negative mean return (-1.65%), and
careful consideration is necessary. Other candidates include Food (0.63) and Machinery (0.89).
Table 40: Comparison between three Portfolio Compositions (Mitsubishi Corporation)
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
15
"Since"all"sectors"and"industries"in"the"market"are"sorted"into"larger"segments"based"on"the"current"organization"
of" Sumitomo" Corporation," the" segmentation" does" not" work" for" these" companies" and" the" precise" examination"
cannot"be"conducted."For"further"discussion,"performance"attribution"for"each"company"would"be"recommended."
16
"Because" of" the" statistically" insignificant" results" of" analysis" made" for" Sumitomo" Corporation," further" analysis" is"
omitted."
73"
"
Segment
IndustrialIFinance,ILogisticsI&IDevelopment
Current
Tangency
Suboptimal
Portfolio
Portfolio
Portfolio
Weights
Weights
Weights
(%)
(%)
(%)
7.78
0.00
0.00
Energy
14.29
0.00
77.44
Metals
32.01
7.12
0.00
Machinery
17.33
0.00
0.00
Chemicals
7.23
2.96
0.00
21.36
89.92
22.56
LivingIEssentials
Table 41: Comparison between Indicators of three Portfolios (Mitsubishi Corporation)
Current
Tangency
Suboptimal
Portfolio
Portfolio
Portfolio
Weights
Weights
Weights
(%)
(%)
(%)
ExpectedEReturnEbasedEonEHistoricalEMean
2.6756
2.4094
4.3650
StandardEDeviation
1.2638
0.2481
1.2644
SharpeERatio
2.1172
9.7115
3.4523
AccountingEBeta
1.3159
0.9658
1.7070
CAPMEforecastedEReturn
2.5588
1.9089
3.2848
Jensen'sEAlpha
0.1168
0.5005
1.0802
TreynorERatio
2.0324
2.4935
2.5565
Figure 32: Systematic Risk-Return Relationship of Each Portfolio (Mitsubishi)
74"
"
Table 42: Comparison between three Portfolio Compositions (Mitsui & Co)
Segment
Iron<&<Steel<Products
Current
Tangency
Suboptimal
Portfolio
Portfolio
Portfolio
Weights
Weights
Weights
(%)
(%)
(%)
7.24
0.00
0.00
Mineral<&<Metal<Resources
15.50
0.00
13.08
Machinery<&<Infastructure<Projects
18.53
64.97
50.54
9.48
6.53
0.00
Energy
24.19
22.85
29.05
Foods<&<Retail
10.55
0.00
0.00
Consumer<Service<&<IT
8.91
5.64
7.33
Logistics<&<Financial<Services
5.59
0.00
0.00
Chemicals
Table 43: Comparison between Indicators of three Portfolios (Mitsui & Co)
Current
Tangency
Suboptimal
Portfolio
Portfolio
Portfolio
Weights
Weights
Weights
(%)
(%)
(%)
ExpectedEReturnEbasedEonEHistoricalEMean
3.1385
2.8163
4.5020
StandardEDeviation
1.1301
0.3884
1.1131
SharpeERatio
2.7773
7.2514
4.0445
AccountingEBeta
1.5467
1.0385
1.6269
CAPMEforecastedEReturn
2.9872
2.0438
3.1361
Jensen'sEAlpha
0.1514
0.7725
1.3659
TreynorERatio
2.0285
2.7109
2.7665
Figure 33: Systematic Risk-Return Relationship of Each Portfolio (Mitsui)
"
75"
"
Table 44: Comparison between three Portfolio Compositions (Itochu Corporation)
Segment
Textile
Current
Tangency
Suboptimal
Portfolio
Portfolio
Portfolio
Weights
Weights
Weights
(%)
(%)
(%)
7.19
39.82
78.50
ICTF&FMachinery
19.57
34.69
1.44
Energy,FMetalsF&FMinerals
30.48
0.00
20.06
Chemicals,FForestFproductsF
&FGeneralFMerchandise
16.24
14.31
0.00
Food
21.56
11.17
0.00
4.96
0.00
0.00
ConstructionF&FRealty,FFinancialFandF
InsuranceFServiecesF&FLogistics
"
Table 45: Comparison between Indicators of three Portfolios (Itochu Corporation)
Current
Tangency
Suboptimal
Portfolio
Portfolio
Portfolio
Weights
Weights
Weights
(%)
(%)
(%)
ExpectedFReturnFbasedFonFHistoricalFMean
2.3305
2.9768
4.7840
StandardFDeviation
1.6471
0.6973
1.6466
SharpeFRatio
1.4149
4.2692
2.9054
AccountingFBeta
1.4766
1.1897
1.8845
CAPMFforecastedFReturn
2.8570
2.3244
3.6143
Jensen'sFAlpha
P0.5265
0.6524
1.1697
TreynorFRatio
1.5775
2.5013
2.5379
"
Figure 34: Systematic Risk-Return Relationship of Each Portfolio (Itochu)
"
76"
"
Optimal-Portfolio-Composition-and-History-of-the-CompanyAside from the common characteristic that all companies have high weight in Energy or
Mineral related segments, there seems to be some sort of correlation between the optimal
portfolio composition of a company and the history of the company.
Sumitomo Corporation’s major constituents of suboptimal portfolio are Metal (46.55%),
Media (27.23%), and Mineral Sources (26.22%). The large weight in Metal corresponds to its
history that Sumitomo has its origin in copper mining and smelting.
For Mitsubishi Corporation, the suboptimal portfolio consists of Energy (77.44%) and
Living Essentials (22.56%).
The Living Essentials’ business includes food, clothing, retail,
general merchandise, and products closely linked to people’s lives. This unit is in line with
commerce, the company’s origin.
Mitsui & Co has Mineral & Metal Resources (13.08%), Machinery & Infrastructure
Projects (50.54%), Energy (29.05%), and Consumer Service & IT (7.33%). Mitsui & Co is the
company with the most diversified suboptimal portfolio. Considering the other three companies
all have high suboptimal weight in Energy segment, there seems to be little correlation between
Mitsui’s suboptimal portfolio composition and its origin, which is rice and coal trading.
As to Itochu Corporation, suboptimal portfolio suggests Textile (78.50%), ICT &
Machinery (1.44%), and Energy Metals & Minerals (20.06%). Itochu used to specialize in textile
trading, and its origin fits to its most favorably weighted business unit in suboptimal portfolio.
The relationship between optimal portfolio and history may be explained by business
units’ positive excess market return and operational stability generated by market power and
expertise acquired through long-term operation in the industry. However, this relationship is not
clear and requires further study.
77"
"
6.5. Implication for the role of Real Estate
The performance of subject companies’ business units that includes the real estate sector
is summarized in Table 46. Except for Sumitomo Corporation, the other three companies all have
very low mean return, high volatility relative to its mean return, low accounting beta, and
negative Jensen’s Alpha. However, as one can imagine, it is difficult to draw a general conclusion
on the role of real estate sector within a corporate portfolio for multiple reasons.
First, since the business model of the subject companies is somewhat Japan-specific and
the population itself is very small, the number of samples is not enough to have conclusive results.
Second, all four companies have real estate section in combination with other businesses,
and none of them has a unit that solely deals in real estate. This organizational characteristic
makes real estate’s contribution to total return and risk of a portfolio more obscure. 17
Last but not least, numerous firm-specific factors affect a firm’s portfolio composition,
which makes generalization a challenge. For these companies, return of each unit can be
interpreted as a cumulative result of project selections including trading or manufacturing. Unlike
security investment, many transactions in industries take place based on private information, and
profitability of a company depends on its market power and competitive advantage in the industry.
In addition, optimal portfolio composition of a firm is determined based on relationships between
business units and overall firm strategy, and hence the relative position of a unit in the firm also
matters to a great degree.
Real estate’s role in a corporate portfolio is different from company to company because
each company’s optimal corporate portfolio greatly varies depending on its organizational
structure, profit structure, and overall strategy.
"""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
17
"For"example,"while"Sumitomo"Corporation"has"unit"accounting"beta"of"0.79,"its"unit"consists"of"real"estate"and"
the"other"two:"food"and"general"products,"which"intuitively"seem"to"have"lower"betas,"and"real"estate’s"accounting"
beta"could"be"higher"than"the"overall"unit"beta."
78"
"
Table 46: Comparison between Business Units including Real Estate
Sumitomo
General3Products
&3Real3Estate
Mitubishi
Mitsui
Construction3&3
Industrial3Finance,3
Consumer3Service3 Realty,3Financial3
Logistics3&3
&3IT
and3Insurance3&3
Development
Logistics
Mean3Return
0.0193
0.0042
J0.0120
J0.0165
Standard3Deviation
0.0067
0.2224
0.3938
0.0431
Sharpe3Ratio
2.8600
0.0188
J0.0306
J0.3825
Accounting3Beta
0.7910
0.5455
0.1933
J0.6846
Expected3Return
0.0158
0.0113
0.0047
J0.0115
Jensen's3Alpha
0.0034
J0.0071
J0.0168
J0.0050
Treynor3Ratio
0.0244
0.0076
J0.0623
0.0241
Average3Correlation
0.5349
0.2641
0.1186
0.2283
13.71
7.78
8.91
4.96
0.00
0.00
7.33
0.00
Portfolio
Weight
Current
Suboptimal
Other3Businesses
Food
Financial3
Services
Services
Finance3&3
Insurance
General3Products
Logistics
Medical3&3
Healthcare
Logistics
within3Unit
Fashion
Housing3&3
Industrial3
Material
79"
"
Itochu
7. Conclusion and Further Discussion
Conclusion-and-LimitationsThis study suggests a framework to optimize a firm’s corporate portfolio using finance
theories and performance evaluation techniques with the consideration of stakeholders’ value
maximization. In the framework, starting from “suboptimal portfolio,” a firm seeks a portfolio
composition which decreases its systematic risk but achieves excess market return at the same
time in order to maximize both shareholders’ and employees’ wealth. When adjusting its asset
allocation, a firm needs to understand how and how much the current corporate portfolio creates
diversification effects and to consider the consequence of the adjustment on the effects.
Although this paper aims to draw a somewhat general conclusion and implications for
real estate sector, the findings of this study are more or less inconclusive for following reasons.
First, all subject companies have different level of competitiveness and profitability in industries
unlike security investment because they face different investment opportunities, so it is extremely
difficult to generalize corporate portfolio management strategy. Rather, corporate portfolio should
be tailored for a firm based on relative position of the firm in industries, relationships between
business units including return correlation and synergy, and overall corporate strategy. Second,
the relationship between diversification effects, such as enhanced return and reduced risk, and
portfolio composition is unclear. Hence, this study cannot incorporate the impact of allocation
adjustment on each unit return and volatility and the results have limited implications.
Further-DiscussionsTo further develop the discussions, additional studies on the relationship between
portfolio weights of business units and diversification effects in terms return and volatility is
recommended. The clarification of this relationship would greatly enhance the framework in this
paper and help the construction of a firm’s overall management strategy as well.
80"
"
8. BibliographyBall,"R.,"&"Brown,"P."(1969)."Portfolio"Theory"and"Accounting."Journal(of(Accounting(Research,(Vol.(7,(
No.(2,"300c323."
Bettis," R." A.," &" Hall," W." K." (1982)." Diversification" Strategy," Accounting" Determined" Risk," and"
Accounting"Determined"Return."The(Academy(of(Management(Journal,(Vol.(25,(No.(2,"254c264."
Bettis," R." A.," &" Mahajan," V." (1985)." Risk/Return" Performance" of" Diversified" Firms." Management(
Science,(Vol.(31,(No.(7,"785c799."
Bildersee," J." S." (1975)." The" Association" between" a" MarketcDetermined" Measure" of" Risk" and"
Alternative"Measure"of"Risk."The(Accounting(Review,(Vol.(50,(No.(1,"81c98."
(2011)."In"Z."Bodie,"A."Kane,"&"A."J."Marcus,"Investments"(pp."846c851)."McGrawcHill"Irwin."
(2011)." Applied" Portfolio" Management." In" Z." Bodie," A." Kane," &" A." J." Marcus," Investments" (p." 822)."
McGrawcHill"Irwin."
(2011)."Index"Models."In"Z."Bodie,"A."Kane,"&"A."J."Marcus,"Investments"(pp."246c274)."McGrawcHill"
Irwins."
(2011)."Multi"Factor"Model:"An"Overview."In"Z."Bodie,"A."Kane,"&"A."J."Marcus,"Investments"(pp."319c
323)."McGrawcHill"Irwin."
(2011)." Optimal" Riske" Portfolios." In" Z." Bodie," A." Kane," &" A." J." Marcus," Investments" (pp." 196c219)."
McGrawcHill"Irwin."
(2011)."The"Capital"Asset"Pricing"Model."In"Z."Bodie,"A."Kane,"&"A."J."Marcus,"Investments"(pp."280c
293)."McGrawcHill"Irwin."
Bowman," R." G." (1979)." The" Theoretical" Relationship" Between" Systematic" Risk" and" Financial"
(Accounting)"Variables."The(Journal(of(Finance,(Vol.(34,(No.(3,"617c630."
(2011)."Portfolio"Theory"and"the"Capital"Asset"Pricing"Model."In"R."A."Brealey,"S."C."Myers,"&"F."Allen,"
Principles(of(Corporate(Finance(Global(Edition"(pp."213c240)."McGrawcHill/Irwin."
(2011)."Should"Firms"Be"Managed"for"Shareholders"of"All"Stakeholders?"In"R."A."Brealey,"S."C."Myers,"
&"F."Allen,"Principles(of(Corporate(Finance"(pp."39c40)."McGrawcHill"Irwin."
Chan," L." K.," Karceski," J.," &" Lodonishok," J." (1998)." The" Risk" and" Return" form" Factors." Journal( of(
Financial(and(Quantitative(Analysis,(Vol.(33,(No.(2,"159c188."
Eisenhauer,"J."G."(2003)."Regression"through"the"Origin."Teaching(Statistics,"76c80."
Elton," E." J.," &" Gruber," M." J." (1997)." Modern" portfolio" theory," 1950" to" date." Jounal( of( Banking( &(
Finance,(Vol.(21,"1743c1759."
81"
"
Encyclopædia" Britannica." (2012)." Mitsui( Group." Retrieved" July" 25," 2012," from" Encyclopedia"
Britannica:"http://www.britannica.com/EBchecked/topic/386291/MitsuicGroup"
Encyclopædia" Britannica." (2012)." Zaibatsu." Retrieved" July" 27," 2012," from" Encyclopædia" Britannica:"
http://www.britannica.com/EBchecked/topic/655416/zaibatsu"
(2007)." Equilibrium" Asset" Valuation" and" Real" Estate's" Price" of" Risk" in" the" Capital" Market." In" D." M."
Geltner,"N."G."Miller,"J."Clayton,"&"P."Eichholtz,"Commercial(Real(Estate(Analysis(&(Investments"(pp."
561c583)."CENGAGE"Learning."
(2007)." PortfoliocLevel" Performance" Attribution." In" D." M." Geltner," N." G." Miller," J." Clayton," &" P."
Eichholtz,"Commercial(Real(Estate(Analysis(&(Investments"(pp."693c698)."CENGAGE"Learning."
Hahn,"G."J."(1977)."Fitting"regression"models"with"no"intercept"term."Journal(of(Quality(Technology,(
9(2),"56c61."
Hamada,"R."S."(1969)."Portfolio"Analysis,"Market"Equilibrium"and"Corporation"Finance."The(Journal(of(
Finance,(Vol.(24,(No.(1,"13c31."
Hamada,"R."S."(1972)."The"Effect"of"the"Firm's"Capital"Structure"on"the"Systematic"Risk"of"Common"
Stocks."The(Journal(of(Finance,(Vol.(27,(No.(2,"435c452."
Henderson," B." D." (1973)." The( Experience( Curve( W( Reviewed( IV.( The( Growth( Share( Matrix( or( The(
Product(Portfolio."The"Boston"Consulting"Group."
Itochu"Corporation."(2011)."Annual(Report(2011.""
Itochu" Corporation." (2012)." Itochu( History." Retrieved" July" 25," 2012," from" Itochu" Corporation:"
http://www.itochu.co.jp/en/about/history/"
Jensen,"M."C."(1967)."The"Performance"Of"Mutual"Funds"In"The"Period"1945c1964."Journal(of(Finance,(
Vol.(23,(No.(2,"389c416."
Lee,"F."C.,"&"Lee,"C."A."(2006)."Encyclopedia(of(Finance."Springer"Science"+"Bisomess"Media."
Lifson,"T."B."(1981)."A"Theoretical"Model"of"Japan's"Sogo"Shosha"(General"Trading"Firms)."Academy(of(
Management(Proceedings,"69c73."
Lintner," J." (1965)." The" Valuation" of" Risk" Assets" and" the" Selection" of" Risky" Investments" in" Stock"
Portfolios"and"Capital"Budgets."The(Review(of(Economics(and(Statistics,(Vol.(47,(No.(1,"13c37."
Lubatkin," M.," &" Rogers," R." C." (1989)." Deversification," Systematic" Risk," and" Shareholder" Return:" A"
Capital"Market"Extension"of"Rumelt's"1974"Study."The(Academy(of(Management(Journal,(Vol.(32,(No.(
2,"454c465."
Markowitz,"H."M."(1952)."Portfolio"Selection."The(Journal(of(Finance,(Vol.(7,(No(1.,"77c91."
Melicher," R." W.," &" Rush," D." F." (1973)." The" Performance" of" Conglomerate" Firms:" Recent" Risk" and"
Return"Experience."The(Journal(of(Finance,(Vol.(28,(No.(2,"381c388."
82"
"
Mitsubishi"Corporation."(2011)."Annual(Report(2011.""
Mitsubishi" Corporation." (2012)." History." Retrieved" July" 25," 2012," from" Mitsubishi" Corporation:"
http://www.mitsubishicorp.com/jp/en/about/history/"
mitsubishi.com" committee." (2012)." About( Mitsubishi:( Origin." Retrieved" July" 25," 2012," from"
mitsubishi.com:"http://www.mitsubishi.com/e/history/index.html"
Mitsui"&"Co."(2011)."Annual(Report(2011.""
Mitsui" &" Co.," Ltd." (2012)." Mitsui's( History." Retrieved" July" 25," 2012," from" Mitsui" &" Co.," Ltd.:"
http://www.mitsui.com/jp/en/ir/annual/history.html"
Mossin,"J."(1966)."Equilibrium"in"a"Capital"Asset"Market."Econometrica,(Vol.(34,(No.(4,"768c783."
Nikkei" Inc." (2012)." High( Market( Cap( Ranking." Retrieved" July" 13," 2012," from" Nikkei.com:"
http://www.nikkei.com/markets/ranking/stock/caphigh.aspx"
Pidun,"U.,"Rubner,"H.,"Kruhler,"M.,"Untiedt,"R.,"The"Boston"Cousulting"Group,"Nippa,"M.,"et"al."(2011)."
Corporate"Portfolio"Management:"Theory"and"Practice."Jounal(of(Applied(Corporate(Finance,(Volume(
23(Number(1,"63c76."
Porter," M." E." (1980)." Industry" Structure" and" Competitive" Strategy:" Keys" to" Profitability." Financial(
Analysts(Journal,(Vol.(36,(No.(4,"30c41."
Sharpe,"W."F."(1964)."Capital"Asset"Prices:"A"Theory"of"Market"Equilibrium"under"Conditions"of"Risk."
The(Journal(of(Finance,(Vol.(19,(No.(3,"425c442."
Sharpe,"W."F."(1966)."Mutual"Fund"Performance."The(Journal(of(Business,(Vol.(39,(No.(1,"119c138."
Smith," K." V.," &" Schreiner," J." C." (1969)." A" Portfolio" Analysis" of" Conglomerate" Diversification." The(
Journal(of(Finance,(Vol.(24,(No.(3,"413c427."
Sumitomo"Corporation."(2005)."Consolidated(annual(results(FY2004.""
Sumitomo"Corporation."(2011)."Annual(Report(2011.""
Sumitomo" Corporation." (2012)." History( of( Sumitomo." Retrieved" July" 25," 2012," from" Sumitomo"
Corporation:"http://www.sumitomocorp.co.jp/english/company/history/index.html"
Treynor,"J."L."(1966)."How"to"Rate"Management"Investment"Funds."Harvard(Business(Review,(Vol.(43."
Untiedt," R.," &" Pidun," U." (2011)." Corporate" Portfolio" Management" Applications:" Role" of" Firm"
Attributes"&"Impact"on"Coporate"Processes."Academy(of(Management(Annual(Meeting(Proceedings.""
YCharts." (2012)." Market( Cap( Rankings."
http://ycharts.com/rankings/market_cap"
83"
"
Retrieved"
July"
26,"
2012,"
from"
YCharts:"